(Colin Campbell M.SC., C.Eng., M.I.Mech - Engrs PDF

Automo bile Suspensions
AutoInobile
Suspensions
COLIN CAMPBELL
M.Sc., C.Eng., M.I.Mech.Engrs
LONDON
CHAPMAN AND HALL
First published 1981
by Chapman and Hall Ltd
11 New Fetter Lane, London EC4P 4EE
© 1981 C. Campbell
Softcover reprint of the hardcover 1st edition 1981
ISBN-13: 978-0-412-16420-0 (cased edition)

ISBN-13: 978-1-4613-3391-3 (paperback edition)
This title is available
in both hardbound and paperback
editions. The paperback edition is
sold subject to the condition that it
shall not, by way of trade or otherwise, be
lent, re-sold, hired out, or otherwise circulated
without the publisher's prior consent in any form of
binding or cover other than that in which it is
published and without a similar condition
including this condition being imposed
on the subsequent purchaser
All rights reserved. No part of
this book may be reprinted, or reproduced
or utilized in any form or by any electronic,
mechanical or other means, now known or hereafter
invented, including photocopying and recording,
or in any information storage or retrieval
system, without permission in writing
from the publisher
British Library Cataloguing in Publication Data
Campbell, Colin
Automo bile suspensions

1. Automobiles - Springs and suspension
I. Title
629.2' 43 TL 257 80-41445
ISBN-13: 978-1-4613-3391-3 e-ISBN-13: 978-1-4613-3389-0
DOl: 10.1007/978-1-4613-3389-0
TO THE MEMORY OF
MAURICE OLLEY
who introduced us to the benefits
of independent suspension
Contents
Preface page ix
1 Wheels and tyres 1
2 Springs 22
3 Suspension principles 41
4 Suspension geometry 54
5 Conventional systems 77
6 Road -holding 105
7 Dampers 120
8 Pneumatic suspensions 129
9 Hydropneumatic suspensions 143
10 Interconnected and no-roll suspensions 165
11 A small FWD saloon car: Ford Fiesta S 185
12 A high-performance sports car: Porsche 928 198
Index 211
We have had enough of action and of motion we,
Roll'd to starboard, roll'd to larboard,
When the surge was seething free.
Alfred Lord Tennyson
Preface
This book is an introduction to the elementary technology of

automobile suspensions. Inevitably steering geometry must
be included in the text, since the dynamic steering behaviour,
road-holding and cornering behaviour are all influenced by
the suspension design. Steering mechanisms and steering
components are not covered in this book.
This is not a mathematical treatise, but only a fool or a
genius would attempt to design a motor vehicle without
mathematics. The mathematics used in this book should
present no problem to a first-year university student. SI units
have been used in general, but for the benefit of those not
familiar with them we have included in brackets, in many
cases, the equivalent values in Imperial units. Many engineers
regard the Pascal as an impractical unit of pressure. The
author has therefore expressed pressures in bars (1
bar = 105Pa). A deviation from SI units is the use of degrees
and minutes, instead of radians, to express camber, castor,
roll angles, etc. This is still common practice in the motor
industry.
No attempt has been made to make any stress calculations
on suspension components. The automobile engineering
student will have access to other textbooks on such subjects
as strength of materials and theory of structures.
The author is grateful for technical information,
photographs and drawings supplied by many car and
component manufacturers. The following companies have
been particularly helpful: Automotive Products Ltd, Citroen
Cars Ltd, Datsun (UK) Ltd, Dunlop Ltd, Lotus Cars Ltd,
Lucas-Girling Ltd, Magnesium Elektron Ltd (MELMAG
wheels), Mercedes-Benz (UK) Ltd, Moulton Developments
Ltd (Hydragas suspension), Porsche Cars (GB) Ltd, Tech Del
Ltd (Minilite wheels).
Special thanks are due to the editor of Motor for his
permission to publish data from 'The 150 m.p.h. Corner', a
x Preface
report (3 March 1979) in that excellent journal, givmg
measurements made with the assistance of the Cranfield
Institute on the Lotus 79 when cornering at 2.05g radial
acceleration.
c.c.
1
Wheels and Tyres
1.1 Suspension
The term suspension suggests 'supporting from above', as in

'suspension bridge' or 'suspender', since the earliest forms of
suspension used on the roads did literally suspend the body
of the horse-drawn carriage at the four corners by very
flexible leaf springs with outer ends that curved over and
upwards to join the shackle pins in the carriage frame.
In the modern automobile suspension plays a dual role. It
'suspends' the body on a shock-absorbing system and at the
same time maintains the contact patches of the four tyres
in effective contact with the road surface. One cannot
overemphasize the importance of this second role. If we fail
in this, the driver cannot control the vehicle. Rally drivers are
well aware that effective road contact is sometimes lost. They
wear safety harnesses and crash helmets to reduce the
dangers from this eventuality and the backup crew carry a
host of spares to repair the damage. Even so, a good
suspension system under normal conditions keeps all four
wheels in contact with the road surface and makes it possible
for a driver of average ability to keep the vehicle under
satisfactory directional control. A bad suspension system fails
to do this under identical road conditions.
It is perfectly logical therefore to begin our analysis of
suspension systems from the bottom, since the car's four
areas of contact with the road surface made by the tyres
(footprints, in American terminology) are so vital to the
behaviour of the vehicle when in motion. This leads us then
to the wheels and the tyres.
1.2 The wheel
Nobody knows who invented the wheel. That delightful

American strip-cartoon series Be, suggested that their most
2 Wheels and tyres
advanced-thinking cave-dweller announced a major
breakthrough when he dropped the idea of a square wheel in
favour of a triangular one. After all, such a wheel would only
give three bumps per revolution instead of four! Somebody
must have progressed towards a crude circle and from a
heavy, clumsy, solid wooden disk, held in place by pegs
driven into holes in the axletree, the wheel had developed
around the year 1800 B.C. into a spoked wheel. The early
Egyptian chariots used only four spokes, but the same
construction principle still survives in our few remaining
wheelwright's shops.
These craftsmen still make a central hub with radial spokes
surrounded by a felloe, or outer rim, made from several
interlocking arcuate pieces. One improvement on the
Egyptian design is the addition of a wrought-iron rim which
is heated in the forge and shrunk in place around the felloe.
These traditional carriage wheels and the wire-spoked steel
wheels that had been developed for the bicycle were the
,alternatives available to the early automobile makers.
1.3 Modem automobile wheels
1. 3.1 Steel wheels

More than 90% of modem wheels are of pressed steel. They
are strong and cheap to manufacture. They require negligible
maintenance and are only inferior to alloy wheels on one
count; they are heavier. Many alloy wheels are bought simply
to impress the neighbours. Even so, they are essential to the
serious rally competitor. All single-seater racing cars use
magnesium alloy wheels. The weights, or more correctly, the
mass of the unsprung components of a vehicle have a
significant influence on the performance of the suspension
system. As will be shown in Chapter 2 the suspension
behaviour is influenced favourably by a reduction in the
unsprung mass. The wheel, of course, is a substantial part of
this mass.
Steel wheels are made from two pressings, as shown in
cross section in Fig. 1.1. The inset distance and the rim
profile can be varied to suit the car manufacturer's
Modern automobile wheels 3
Ventilation
Wheel stud
Nominal or
mting
Pitch diameter
circle
diameter
Protuberance ----;~b1
~ Rim width ---l

Fig. 1.1 Cross section of a steel wheel.
requirements. The flange profile, indicated by letters K, JK, J

or -C in the specification, is designed to comply with the tyre
bead profile.
1;3.2 Aluminium alloy wheels

Aluminium alloy wheels are usually made from heat-treated
castings. A typical example, the Minilite Sports wheel made
by Tech Del Ltd of London, is shown in Fig. 1.2. The
weight-saving compared, with a steel wheel varies from about
30-50% the saving being greater for the wider wheels. The
Minilite wheel is stove-enamelled to improve appearance and
to provide corrosion protection.
4 Wheels and tyres
Fig. 1.2 Minilite aluminium alloy wheel.
1.3.3 Magnesium alloy wheels

A typical aluminium alloy has a density of about one-third
that of steel. Cast magnesium alloys are even lighter, having a
density of slightly less than one-quarter that of steel. The
tensile strength is 40-50% of a typical steel pressing, giving a
great improvement in the specific strength (tensile strength per
unit density). A cast magnesium alloy wheel of 5.5-in rim
width and 13 in nominal diameter, as used on a typical works
rally car weighs 4.2 kg in Minilite form. The weight of a steel
wheel of this size can vary between 6.0 and 6.7 kg. Larger
wheels yield a greater weight-saving. A 6JK x 15 steel wheel
for a Jaguar weights 10.3 kg. A Minilite magnesium
replacement weighs only 6.7 kg.
Rally wheels are designed to withstand very high shock
loads. Since tyres are often ripped open by jagged rocks, the
wheel rims must also suffer the same brutal treatment. Racing
drivers do occasionally clip a kerb, particularly on circuits like
Monaco. In general, however, wheels on racing cars do not
suffer such high impact loads as those fitted to rally cars.
Whereas a rim thickness of 6.4 mm will be used on a rally
magnesium wheel, a single-seater racing car wheel is
Modern automobile wheels 5
13.
Tyrrell Fl McRae FSOOO Lotus Fl Sur t ees F2
13" • 16"reor 15" • 17 '" reor 10" front 13 " ,10" front
Wt 4·10 kg Wt 5·44 kg wt 3·34 kg WI 3-40 kg
Shadow Fl F3 Rondel F2 Brabham F3

13" x l1"front 13 " x Sfrant
" 13 II' x l,d."reor 13"x 10' reor
Wt 3·63kg Wt 2·95 kg Wt 3'SO kg Wt 3·46 kg
Fig. 1.3 Cross sections of several MEL MAG magnesium alloy

racing wheels.
expected to survive with a rim thickness of no more than

2 mm. As with all components in a racing car, a high
standard of reliability can-never be achieved if the car is -to
stand any chance of winning.
Many Grand Prix cars today are fitted with the MELMAG
wheel, invented by Gerry Watts of Magnesium Elektron Ltd
of Swinton, Manchester. Cross sections of several MEL MAG
designs from Formula 3 to Formula 5000 are given in Fig.
1. 3. The MELMAG wheel consists of two deep pressings in
magnesium alloy ZM21. A disk of honeycomb foil, mounted
on the central magnesium alloy hub attachment, is used as a
spacer element. A tubular tension strap envelopes the central
honeycomb spacer. The complete assembly is bonded
together by a high-temperature adhesive. This sandwich
construction of the wheel disk, in which a honeycomb spacer
is used to separate two stressed outer skins, is an aerospace
technique.
1. 3.4 Wire wheels

The centre-lock wire wheel (see Fig. 1.4) is traditionally
associated with vintage sports cars and racing cars, and for
6 Wheels and tyres
Fig. 1.4 Centre lock wire wheels with octagonal hub caps.
those of us of advancing years the blood is still stirred by

memories of split-seconds saved by the deft application of
copper-headed hammers to eared hubcaps. We also
remember the problems of fretting corrosion that could occur
on the conical seatings and the continual attention demanded
by the wire spokes. Only Ettore Bugatti ever made a wire
wheel that was substantially lighter than a good design of
steel wheel and for this he used a very large number of
spokes of fine-gauge, high-tensile piano wire. The excellent
brake-cooling afforded by wire wheels made them
irreplaceable in the 1930s. Even with the 1937 Formula 1
Mercedes that transmitted about 600 b.h. p. to the road
surface, wire wheels were still reliable when serviced with
Teutonic regularity.
Centre-lock wire wheels are now only fitted as an 'optional
extra' on a few sports cars. The weight-saving is negligible.
The Morgan 4/4, for example, can be supplied with 5 x 15
centre-lock wire wheels of 8.75 kg weight. A replacement
steel wheel would be no heavier, but the aesthetic appeal is
undeniable and the type of driver who buys a Morgan
usually prefers wire wheels.
The Wheel Division of Dunlop Ltd also supply
chrome-plated bolt-on wire wheels. These again offer no
saving in weight, but are a direct replacement for the
standard 4- or 5-stud steel wheel. They are optional
equipment on such sports and 'sporting' cars as the Datsun
280Z, the Opel Manta and the Ford Mustang.
The tyre 7
'Elastic belt' 01
rubberized canvas
Fig. 1.5 Thomson's original pneumatic tyre of 1845.
1.4 The tyre;
When the process of vulcanizing natural rubber became a

commercial proposition (albeit a very poor product by modem
standards), it was no great step to replace the steel-rimmed
wheel by one carrying a solid rubber tyre. I myself have
painful memories of solid tyres, still in use on commercial
vehicles when I entered my teens. At 13 I drove a 4-ton
Leyland lorry (illegally, of course) with cast-iron spoked
wheels fitted with solid rubber tyres. The shocks transmitted
up by the steering column by the Lancashire stone setts at
12 m. p.h. (the legal speed limit for a lorry) were almost as
bad as the vibrations of a pneumatic road hammer. Even so,
it was not the idea of comfort that led John Boyd Dunlop to
make pneumatic tyres for the tricycle ridden by his
10-year-old son. He noted what a struggle the lad had to
pedal his tricycle along a muddy lane. Fat-section pneumatic
tyres, he reasoned, would not sink so readily into the soft
mud and would not require as much physical effort from the
rider. Dunlop made his first pneumatic tyre in 1888 and
remained unaware for many years, even after a company had
been formed to manufacture his tyres, that Robert William
8 Wheels and tyres
Thomson had patented a pneumatic tyre in 1845 (see Fig.
1.5) and had even made measurements of the reduced
tractive effort required to pull a light brougham over a paved
road, a macadam road and a surface of crushed granite, when
a change was made from wrought-iron tyres to pneumatic
ones.
There was little interest in Thomson's inve,ntion at the
time. No doubt horses would have appreciated this
improvement, but horses are not as articulate as cyclists and
tyres did not appear on the scene in any numbers until the
end of the nineteenth century. Cyclists very soon established
superior ride and reduced road resistance of the products
marketed by the Pneumatic Tyre and Booth's Cycle Agency
Ltd, of which J. B. Dunlop was a director. Since Dunlop's
company* did not have a monopoly, there were many people
working on the problems associated with these most
unreliable early pneumatic tyres. It was the Michelin Co., in
France, who were the first to market a pneumatic tyre for the
new horseless carriages.
Fig. 1.6 C. K. Welch's rim and bead design of 1890.
'Later reregistered as Dunlop Pneumatic Tyre Co.

The tyre 9
Fig. 1.7 William Bartlett's rim and bead design of 1890.
Two inventions that improved the reliability and

convenience of the pneumatic tyre were registered in 1890.
An Englishman, C. K. Welch, invented a detachable tyre
incorporating high-tensile steel wires in what we now call the
'beaded edge' (see Fig. 1.6). Only 36 days later an American,
William Bartlett, filed an invention (Fig. 1.7) with an
improved beaded edge that was locked in position on the rim
by the internal air pressure. Bartlett did not use steel wire in
his beaded edge. By combining the two systems, we arrive at
the beaded-edge construction used in 99% of modern tyres.
The internal construction of a modern cross-ply (bias-ply in
the USA) tyre is shown in Fig. 1. 8.
1.4.1 Tyre construction

The early tyre makers decided to strengthen the walls of the
tyre with a cotton -canvas reinforcement. This seemed a
sensible decision at the time but it proved to be a disastrous
mistake. By using both warp and weft interwoven in this
manner, a chafing action resulted as the tyre section flexed at
the road contact patch. This generated heat in the natural
rubber compound and led to a rapid breakdown in the
10 Wheels and tyres
Fig. 1.8 Internal construction of cross-ply tyre.
reinforcement. The average life of the early car tyre was about
2000 miles.
In the 1920s the cross-ply construction was developed in
which the tyre casing was built up by placing several plies of
parallel cords inside the mould, the cords passing at a bias
angle from the inside beaded edge to the outer beaded edge.
Alternate plies, however, had the bias angle (relative to the
circumferential centre line) alternate between a positive and a
negative angle. The all-important difference from the earlier
canvas reinforcement is that each separate sheet of ply has
parallel cords which are coated completely with a rubber
compound. Direct contact between adjacent cords is,
therefore, avoided. The alternating angles between the bias
plies and the circumferential centre line is usually between
20° and 30°. Large angles are chosen when comfort is the
primary consideration. Racing tyres are made with an angle of
about 20°, giving a harsh ride, but increased resistance to the
centrifugal forces associated with racing speeds.
Until 1948 when Michelin introduced the 'X' series radial
tyres, many tyre manufacturers believed there was little left
for their research-and-development departments to do but to
make minor refinements in construction techniques and to
70 series
60 series
SO series
Fig. 1.9 Tyre footprints.

12 Wheels and tyres
improve the grip and wear resistance of the tread compounds
and patterns. They were well aware that the contact patch (or
footprint) (see Fig. 1.9) suffers a distortion and a scrubbing
action under cornering forces and, to a less extent, under
acceleration and braking. Tread patterns and rubber
compounds were developed to resist this action. In the new
'X' tyres Michelin changed the method of tyre construction to
reduce the amount of distortion in the footprint zone and
immediately doubled the life of car tyres using identical
compounds. Michelin stiffened the zone behind the tread by
bands of steel mesh. They also increased the flexibility of the
side walls by increasing the cord bias angle to 90°; hence the
name 'radial', since the textile cords passed radially from
inner to outer beaded edge.
Since the Michelin patents expired many variants on the
theme have emerged, but the common factor is the use of
'" DUNLOP SP SPORT
Fig. 1.10 Internal construction of steel-braced radial tyre.

The tyre 13
radial cords. Some manufacturers still use steel mesh to
stabilize the footprint, some .use nylon belts sandwiched
between steel belts, others use belts made only of synthetic
textiles. A modern example is shown in Fig. 1.10.
A characteristic of the original 'X' tyre was a superior
cornering power that had a tendency to fall rapidly as the
limiting cornering speed was reached. With modern radial
tyres the breakaway is more gradual and well within the
control of the average driver.
With normal road speeds and moderate cornering modern
,radial tyres give mileages of 40-50 000 miles. This has been
something of a mixed blessing to the tyre makers!
1.4.2 Skidding
In the 1920s when tread patterns of the type shown in Fig.
1.11 were in vogue, it was known that ribbed tyres gave long
life while block patterns gave good resistance to skidding.
Until makers like Dunlop developed their Cornering Force
Fig. 1.11 Tread patterns of the 1920s.

14 Wheels and tyres
1.0
>-
z
....
-
<..>
......
....
0
<..>
t:J
z
-0
-...J
'"
SLIDING VELOCITY (MPH)

Fig. 1.12 Effect of sliding velocity on coefficient of friction.
Machine in the 1950s tread patterns were developed and

tested by a loosely co-ordinated system of road-testing and
feedback of information from the customer. It was 1958
before the 1st International Skid Prevention Conference was
held, and 1964 before the American Society for Testing
Materials standardized a Standard Pavement Skid Test Tire in
their specification ASTM-E17.
Early experiments using this tyre soon gave scientific proof
of the obvious; the type and condition of the road surface is
just as important as the tyre. On wet road surfaces
(pavements, in the USA) the coefficient of friction could vary
from almost zero to 0.7. On dry roads the variation was less
drastic, being from 0.4 to 0.87. It will be seen from Fig. 1.12
that the coefficient falls with increase of speed. The Road
Research Laboratory in the UK had already measured
locked-wheel stopping distances on good road surfaces, such
as dry clean concrete or hot-rolled asphalt with precoated
chippings, of 30-40 ft from 30 m. p.h. This represents a
coefficient of friction of about 0.8. One of the poor surfaces
gave a stopping distance from 30 m. p.h. with locked wheels
and in wet conditions of 495 ft. The coefficient of friction
in this case would be very close to zero, confirming the
measurements made on the Standard Pavement Skid-test
Tire.
It is interesting to note that smooth tyres gave almost as
good frictional coefficients as those with tread patterns when
v = 46 m.p.h. V = 79 m.p.h.
SMOOTH TREAD
V = 46 m.p.h. v = 80 m.p.h.
4· GROOVE RIB TREAD
Fig. 1.13 Aquaplaning: water depth, 0.5 mm (photographed from below through transparent road surface).
16 Wheels and tyres
tested on good dry surfaces. To achieve maximum footprint
area, dry racing tyres are completely bald. They are, of
course, extremely dangerous on a wet road, since they
behave like surfboards. At speeds above about 50 m.p.h. the
buildup of water in front of the footprint acts like a wedge to
force a film of water across the whole area of the footprint,
'thus destroying the grip on the road. Ihis phenomenon is
known as aquaplaning (hydraplaning, in the USA). All good
modem tread patterns are designed to allow large volumes of
water to pass from front to rear of the footprint. The
influence of good footprint drainage is clearly illustrated in
Fig. 1.13.
1.4.3 Dynamic footprint behaviour

Skidding, as discussed so far, is complete loss of directional
control, but this does not mean that the tyres travel over the
road surface at all other times like the flanged wheels of a
train on a track. When torque is applied to accelerate or brake
the vehicle, or when centrifugal force is resisted in a comer,
slip always occurs between the tyre and the road surface. Fig.
1.14 shows how a wheel operates at a slip angle to create a
Effective Swivel Point

projected from King-pin
Contact
Fig. 1.14 Distortion of tyre as it approaches footprint zone (extent

of distortion exaggerated),
The tyre 17
cornering force. The slip angle increases if more cornering
force is demanded. This can increase until a point is
eventually reached where an increase in slip angle produces
no increase in cornering force. This is the limit. It occurs at a
slip angle between 10° and 14°, depending upon the
particular tyre design and, of course, the road surface.
Beyond this limit the cornering force decreases with increase
in slip angle and, if the driver makes no steering correction to
reduce the centrifugal force, the tyre begins to slide. This is a
true skid.
The side thrust exerted by the footprint also varies with the
load carried by the tyre, since the size of the contact patch
varies with load. For every value of slip angle there is an
optimum load and an optimum tyre pressure. A typical family
of curves measured by a cornering force machine is shown in
Fig. 1.15. With a change in tyre pressure, a different family of
curves would result. Modern low-profile tyres are very
3·0 500
2·8 400
2·6
2·4
300
~ 2·2
=
z
Q)
2·0
1·8
()
J...
.2 1· 6 200 kg
01
c
·c 1·4
Q)
cJ... 1·2
0
u '·0
0·8
0·6
0·4
0·2
1° 2° 3° 4° 5° 6° 7° SO 9° 10° 1,° 12°

Slip angles
Fig. 1.15 Typical cornering force curves for modern radial.
18 Wheels and tyres
sensitive to tyre pressure. If the pressure is too high, the
footprint area is reduced. If the pressure is below the
optimum, the footprint becomes concave. More load is carried
by the outside zone of the footprint, less by the central zone.
The right-hand 60 series and 50 series footprints shown in
Fig. 1. 9 appear to be slightly underinflated.
1.4.4 Cornering power

Cornering power is defined as cornering force divided by the
slip angle required to generate the force. It is an important
parameter to all tyre designers, but is difficult to show in
graphical form since the family of curves usually falls so close
together. If we study Fig. LIS, we see that this particular
tyre at a load of 300 kg has a cornering power of
0.31 kN/degree at a slip angle of 2 degrees; 0.305 kN/degree
at 4 degrees; 0.275 kN/degree at 6 degrees; and 0.245
kN/degree at 8 degrees. This is typical, with the cornering
power falling with increase in slip angle.
We have already mentioned how tyre pressures and vertical
load can influence cornering forces, and by inference
cornering power. The following factors also influence
cornering power:
Tyre profile. The tyre profile ratio is usually expressed as an

aspect ratio, this being the ratio of height to width of the tyre
cross section. It is variously described by the tyre makers as
'70 profile' or '70 series' when the height is 70% of the width.
Thus a 225/50 VR16 marking on a tyre indicates a tyre width
of 225 mm with a 50 profile on a 16-in diameter wheel and a
VR speed rating (up to 150 m. p.h.). The mixed units are an
indication that metrication is a slow process of integration
across many industries. About twenty-five years ago tyre
profiles had only moved from 100 to 85. The actual profile
(not profile ratio) of the footprint at this time was convex to
the road surface. Measurements of cornering power showed
that a small increase in tyre pressure above the maker's
recommended pressure would give an increase in cornering
power. Modern low-profile tyres present a flat profile to the
road surface when at the designed load and at the
recommended pressure. Very high cornering powers are
given by those with the more extreme aspect ratios and it is
The tyre 19
almost impossible for cautious, elderly motorists like the
present writer to make a car like the Porsche type 928 lose its
grip on a good dry surface. The use of very wide tyres,
however, does demand certain restrictions in the suspension
system, since a change in camber angle of more than two or
three degrees will lift the tread away from the road surface
on one side and seriously reduce cornering power. Racing
cars use aspect ratios as low as 35% to achieve quite
phenomenal cornering powers. Such tyres are very sensitive
to camber changes and tyre pressure and are a source of
considerable worry to both driver and tyre technician.
Rim width. There is an optimum wheel rim width for any tyre
section. If a change is made to tyres with a lower aspect ratio,
it is advisable to change the wheels, too, since the use of
wide tyres on narrow rims introduces the danger of
destructive stresses in the side walls.
Camber. Wheel camber is the angle between the plane of

rotation of the wheel and the vertical. It is conventionally
l f8 iI J
L I
J
Cornering
777~T~~~//~ force
Positive camber
,
I~ i l~
L i ,
Cornering
7777J»1!f17i77777777777777777? ~ force
Negative camber
Fig. 1.16 Camber.

20 Wheels and tyres.
defined as positive when the top of the wheel leans towards
the cornering side force. This is illustrated in Fig. 1.16. As
already stated, with tyres of aspect ratio greater than 80% an
increase in cornering power is given when a wheel is run at a
negative camber. By extending the tread pattern well into the
side walls, racing motor cycles gain the ability to comer at
very high speeds, using what appear to be impossibly high
negative cambers. It is interesting to compare this with the
contrasting development in tyres for racing cars where
ultra-low aspect ratios demand the virtual elimination of
camber.
Load. Every tyre is designed for a specified load. There is a

certain flexibility in this parameter and the tyre pressure is
usually increased to restore the normal footprint when the
car is heavily loaded. A large increase above the recom-
mended load, however, increases operating tempera-
tures. The safe operating speed is thus reduced on a heavily
loaded car.
Another aspect of load variation of interest to the designers
of high-performance cars is that excessive load transfer when
cornering at high slip angles reduces the ultimate cornering
power. For example, if we take our cornering force values
from Fig. 1.15, we find that a car designed with negligible
load transfer and with both inner and outer tyres loaded to
300 kg, would exert cornering forces of 1. 97 kN from each
tyre at slip angles of 8°. If however the suspension geometry
permits roll to take place to the extent that the inner wheel
load is only 200 kg and the outer is increased to 400 kg,
greater slip angles will be required to exert the required total
cornering force of 3.92 kN. A value of 9.1 degrees would give
about 2.59 kN cornering force on the outer wheel and
1.38 kN on the inner. Load transfer, therefore, reduces
cornering power.
Traction, braking and acceleration. It has been shown

experimentally that with few exceptions a tyre footprint will
exert the same limiting force in any direction. Under braking or
acceleration, the limiting cornering forces are therefore
reduced. A close approximation to the limiting cornering force
in such cases is given by a vectorial combination of the two
The tyre 21
CONTACT
PATCH
CIRC LE DEFINES LI M ITING

TYRE ADHESION FORCES
.-"-f-------~/
CORNERING ONLY
CORNERING +TRACTION
CORNERING + BRAKING
Fig. 1.17 Circle of Forces: how available cornering force is reduced

during traction and braking.
forces (cornering plus traction, or cornering plus braking).

The circle of forces shown in Fig. 1.17 demonstrates this.
Reference
[1] Setright, L. J. K. (1972), Automobile Tyres, Chapman & Hall,

London.
2
Springs
2.1 Comfort and fatigue
The degree of discomfort we are prepared to accept from an

automobile suspension depends largely on the vehicle and its
purpose. A racing driver will endure vibrations that blur his
vision and vertical accelerations that jar his spine, yet he only
mentions such things to the team manager if he has noticed a
su btle change from the norm, since such a change could
indicate a flat spot on a tyre or an impending transmission
failure. This same driver, however, will complain that his
new sports car has a hard ride, even though by comparison
with his racing car the sports car gives a ride like a
feather-bed. In a sports car most drivers will accept a
relatively hard ride as a necessary adjunct to precise
high-speed handling and safe fast cornering. In a large saloon
a similar ride would be considered intolerable.
22 Tolerance to vibration
The medical profession has been interested in the effects of

vibration on the human body for many years. In particular,
studies have been made to establish which particular
frequencies are harmful and produce fatigue quickly in
industrial workers. Raynaud's phenomenon, for example, is
associated with high-frequency vibrations in the range
20-200 Hz from handheld power tools, while motion sickness
is more likely to occur with very low-frequency vibrations in
the range 0.1-0.5 Hz. As a broad guide, we can say that the
human frame experiences fatigue most quickly when exposed
to vibrations in the range 4-8 Hz. A working group of
Technical Committee 108 of the International Organization
for Standardization gave recommendations for the time-
limits beyond which one can expect a worker to show signs
Spring design 23
of 'fatigue-decreased proficiency'. Their recommended
time-limits for vibrations at an RMS acceleration level of
1.0 m/s 2 (approx. 0.1 gravity) were 1.5 h at a frequency band
of 4-8 Hz, increasing to 4 h at 1.0 Hz. Obviously we should
aim to provide a suspension system that confines the major
vibration frequencies very close to 1.0 Hz, avoiding if possible
any prolonged subjection of the occupants to frequencies as
low as 0.5 Hz.
2.3 Spring design
2.3.1 The simple undamped spring
A simple spring system is shown in Fig. 2.1. In the mean

position the force exerted by the coil spring P exactly balances
the weight (or gravitational force) W of the mass M. If the
mass is displaced by distance x, the spring force will increase
to P + kx, where k is the stiffness or rate of the spring, i.e. the
force per unit displacement.
M,ofmoo_- p
X 1-----
o;,~o~J pos",o,__ L____._
Fig. 2.1 Simple spring system.

24 Springs
I
X
--~~-----+--~Y4-~-+-L----~--------~~~
Fig. 2.2 Simple Harmonic Motion.
In the displaced position the body of mass M, if released,

will accelerate towards the mean position at a value of x.
Since x is considered positive in a downward direction, this
acceleration will also be positive. Conversely, if the body is
displaced in an upward direction by an amount -x, the
acceleration will be in the opposite direction. This system,
therefore, constitutes a Simple Harmonic Motion and the
periodic time for such a motion is:
T = ~ = 21tJ~ . (2.1)
OJ is the angular speed in Circular Harmonic Motion, as

Spring design 25
illustrated in Fig. 2.2. This equation can be expressed in
. . . - 21l'J
words as:
Mass of body (2.2)
Penodlc tIme - S· ffn f . .
tI ess 0 sprmg
Alternatively, if we take the deflection of the spring under
the weight of the body as d (the static deflection):
T = 21l'Ji (2.3)
where g, in this case, is the acceleration due to gravity.
2.3.2 Body accelerations

Until road authorities provide road surfaces as smooth as a
billiard-table, automobiles will continue to need springs. Let
us take the simple case of a car travelling at speed that
encounters a 50-mm step in the road level, the kind of step
one could easily encounter when a new road surface is being
laid. For simplicity we will neglect the spring in the tyres, the
behaviour of the spring dampers and the springs in the seats.
The sudden compression of the front springs of the car by a
50-mm step in the road will not produce an instant lift of
50 mm in the car body. This would be a shattering experience
to both car and occupants. What actually happens is a
compression of the two front springs by 50 mm. The springs
then begin to release this stored energy by lifting the front
end of the body. The rate at which this occurs is controlled
by the natural frequency of the springs. The maximum
acceleration transmitted to the body is given by a
rearrangement of the Equation (2.3):
21l')
a =d (T
2
(2.4)
Taking a typical front spring frequency of 1.25 Hz, i.e.

T = 0.8 s:
21l')2
a = 0.05 x ( 0.8
= 3.08 m/s 2
= 0.31 gravity.
26 Springs
r-~~ ~~.=~___.~
__ ._._J_ =t~Jl_m__m
__ : ______
Fig. 2.3
This maximum acceleration to which the occupants will be

subjected will occur at a quarter-oscillation of the springs after
the front wheels strike the 50-mm step, i.e. after 0.2 s. If the
car is travelling at 50 m.p.h. (22.4 m/s), the occupants will
.feel the maximum acceleration about 4.5 m after the wheels
strike the step. With the undamped springs the initial
compression of 50 mm will lift the front end of the body a
total distance of 100 mm, since the springs will overshoot the
equilibrium position by another 50 mm (see Fig. 2.3).
Effective damping is, therefore, essential if the car is not to
proceed along the road like a yo-yo. For a given spring rate,
good damping does however carry the penalty of an increase
in maximum acceleration. On the other hand, good damping
does permit the use of a softer suspension and the overall
effect is most certainly beneficial.
If we doubled the spring rate of the example above, i.e.
using springs with a frequency of 2.5 Hz, as one might find
on a vintage sports car, the maximum acceleration would
become:
a = 0.05 x (-21t) 2
0.4
= 12.4 m/s 2
= 1.26 gravity.
With high-pressure narrow tyres and very little help from

vintage-type upholstery, the inevitable result would be that
the occupants would momentarily lose contact with their
Spring design 27
seats. This explains why drivers at Brooklands used body
belts to support their abdominal muscles against the
hammering inflicted by the irregularities of the concrete oval.
2.3.3 Spring rates

Springs can take a multitude of shapes: parallel coils, conical
coils, torsion bars, spirals, disks, single leafs, laminated leafs;
there seems to be no end to the possibilities. One need not
be confined to metallic materials. They can be made of rubber
or take the form of a gas container compressed by a piston or
diaphragm. Today the most popular spring in use on
automo biles is the coil spring with coils of identical diameter
and spacing. It is also convenient and certainly inexpensive to
use a constant gauge from top to bottom. Such a spring
usually gives a close approximation to a constant spring rate
over its working range. If we compress it 10 mm and it exerts
a force of F Newtons, compression to 100 mm will give a
restoring force of 10F Newtons. One can, of course, vary the
spring gauge or the diameter of the coils to give a
variable-rate spring, but our mathematical analysis would, at
'this stage of our study, become very long and tedious if w~
introduced such complexity.
There are certain physical limitations to the use of very low
spring rates. The lower the spring rate the greater the initial
deflection (the static deflection) under the weight of the body
and, of equal importance, the greater will be the total travel
--
.
'"
.........
\ \ \
'-... . _ . - '- - -- -.-..... "- ' - ' -
"-
\ ""'"
\
\
./ )
Stotic Bump Rebound
Fig. 2.4 Bump and rebound.

28 Springs'
of the wheel in bump and rebound (jounce, in the USA) (see
Fig. 2.4).
Let us consider a case for which the sprung mass M, is
1200 kg, which with a 50/50 weight distribution front to rear
gives a load of 300 kg per wheel. From Equation (2.1), we
derive:
k=~ (~r
or, since T = 1// (where / is the frequency in Hz):
M
k= "4 (27Cf)2 N/m. (2.5)
Table 2.1 demonstrates how an increase in spring natural

frequency increases the spring rate and decreases the static
deflection. This table helps to highlight our problem. If we
want a comfortable ride, the spring frequency must be low,
preferably below 1.5 Hz. To achieve this, the suspension
system and its attendant linkages must permit large wheel
movements, often as high as 250 mm (10 in). When modern
tyre designs with low-profile ratios demand suspension
systems that give negligible change in camber angle, we
begin to see how hedged around with conflicting
requirements we have become. It will take several chapters
before we are able to show how a few contemporary
suspension engineers have achieved satisfactory compromises
between these conflicting demands.
Table 2.1 Spring rates and static deflections under 300 kg sprung mass
Spring rate k Static deflection d
Frequency f
(Hz) (N/mm) Obf/in) (mm) (in)
1.00 11.8 67 248 9.8

1.25 18.5 105 159 6.3
1.50 26.7 151 110 4.3
1.75 36.3 206 81 3.2
2.00 47.3 269 62 2.4
2.50 74.0 422 40 1.6
Wheel contact 29
2.4 Wheel contact
Wheel contact with the road surface is obviously desirable at

all times, yet we know from experience that it is not all that
uncommon for one or even both wheels at front or rear of the
car to leave the ground. In some of the more spectacular
rallies the cars only seem to spend about half the time with
all four wheels on the ground.
What are the factors, then, that help us to maintain the
tyres in contact with the ground? For simplicity, the ripple is
considered to be sinusoidal and the movements of the two
front wheels to b~ in phase. Again to simplify matters the
mass of the front part of the body, 2Mf is considered to be
capable of independent movement in relation to the rear. We
can, therefore, treat a single front suspension system as if it
were a pogostick with a single mass M f mounted on one
wheel (see Fig. 2.5). The unsprung mass, i.e. the front
wheel, approximately one-half of the coil spring and one-half
the mass of the suspension links, is mf.
Let us take the case of a spring of frequency f = 1.5 Hz and
a ripple depth x = 150 mm. From Table 2.1, we see this
depth exceeds the static deflection of the spring (110 mm).
The tyre contact patch will, therefore, be completely unloaded
at the bottom of the ripple. The unsprung mass has inertia
and will cause the wheel to overshoot this point, but not by
t""1"~---- S - - - - - - 1__-',
Fig. 2.5 Loss of contact in a ripple.
30 Springs
as much as 40 mm. Therefore, we can predict that a ripple
depth of about 150 mm is the limit for such a spring system,
if wheel contact is to be maintained.
If we soften the spring suspension to a frequency of
1.25 Hz, the static spring deflection is increased to 159 mm.
With this new spring the tyre will remain in full contact when
traversing the length of the ripple, even though the load on
the footprint will be reduced at the bottom of the ripple.
We have shown that our pogo stick system can be designed
to maintain reasonable surface contact with a ripple depth of
150 rom (6 in) with a relatively low frequency. There must,
however, be a lower limit to the length of ripple that can be
traversed at a given speed. In this analysis it is assumed that
the car is travelling at such a high speed that the inertia of the
relatively massive sprung body prevents any appreciable
vertical movement. The conditions leading to resonance
bouncing of the body will be discussed later. The natural
periodicity of the spring under the action of the unsprung
mass is not the same as that under the action of the sprung
mass. From Equation (2.1), we see that T ex: ,jM. Thus, with a
natural period of T under mass M f , t = T,j(m/M f ) when the
spring is oscillating under the smaller mass mf'
If we take the limiting ripple length as s metres and the car
velocity as v metres per second, the limiting ripple length is
given by:
Therefore
(2.6)
where f is the natural frequency in Hertz under sprung mass

M f•
Table 2.2 has been calculated for a range of values of m(M
and for a range" of spring frequencies. The speed in all cases is
100 kp.h. (62 m.p.h.).
Movement of the sprung mass 31
Table 2.2 Limiting ripple length for road contact at 100 k.p.h.
Limiting ripple length for road contact at

100 k.p.h. (62 m.p.h.) (m)
Spring frequency f
(Hz) m/M = 1/4 m/M = 1/6 m/M = 1/10
1.0 13.9 11.4 8.8
1.5 9.2 7.6 5.9
2.0 6.9 5.7 4.4
2.5 5.6 4.5 3.5
A close study of Tables 2.1 and 2.2 shows that a low spring
frequency helps the wheels to follow the contours of road
ripples at speeds of 100 k.p.h., but only if the ripples are
spaced at a greater distance than 9-14 m.
On the other hand, the use of a high spring frequency,
such as 2.5 Hz, will reduce these limits to 3.5-5.5 m,
depending upon the unsprung-to-sprung mass ratio, but the
limiting depth of the ripple will be correspondingly reduce~.
The choice of spring frequency is, therefore, not easy. Once
again, we are faced with that unattractive word 'compromise'.
Many modem designers are turning to rising rate suspensions
in an attempt to conquer this problem. Such spdngs give low
frequencies under small wheel movements and higher
frequencies under large movements.
2.5 Movement of the sprung mass
Keeping the wheels on the ground is one design

consideration, but no less important in a passenger-carrying
vehicle, is the movement of the sprung mass, particularly
since you or I, or even our mother-in-law, could be part of it.
Returning to our pogo stick concept, we can apply Newton's
Second Law of Motion. The force imparted by the wheel to
the base of the spring will thus produce an acceleration of the
sprung mass. Since this force is also related to the mass and
acceleration of the unsprung mass, we can state that:
F=ma=MA
32 Springs
where a = the acceleration of the unsprung mass, and
A = the acceleration of the sprung mass.
As a general case, we can state that any road disturbance
producing an acceleration a in the unsprung mass, will
transmit an acceleration in the sprung mass of value
A = a(mfM). For maximum comfort and, of course, to reduce
the shock loads on the body/chassis components to a
minimum, the ratio mfM must be as low as possible.
This general case applies to the typical road contour
composed of random surface irregularities. The case of a
regular series of undulations, such as one meets on unmade
roads in undeveloped countries or on the 'washboard' dirt
roads in the less-populated areas in the USA, is a special
case that calls for special treatment. The evenly spaced ridges
in a washboard road are produced by the natural frequency
of the sus~nsion system, since the driver adopts a speed that
produces the least body movement. In theory a car travelling
at the optimum speed with identical frequencies at both ends
could float along with negligible vertical accelerations
transmitted to the passengers. In practice a typical car would
have a mean frequency of about 1.4 Hz (1.3 Hz at the front
and 1.5 Hz at the rear). If we take this mean value of 1.4 Hz
and a mean of m/M of 1/10, the most comfortable value of v
can be calculated from Equation (2.6):
s = lJ:.
Therefore
v =SfJ~.
If s is taken as 5 m:
v = 5 x 1.4J10
= 22 m/s
= 80 k.p.h. (49 m.p.h.).
A car with a smaller mfM ratio, say 1/12, and with the same
mean spring frequency, would travel most comfortably at a
speed of about 87 k. p.h. (54 m. p.h.).
Vintage sports cars with mean spring frequencies of about
2.0 Hz or even higher would be at a great disadvantage in
Movement of the sprung mass 33
such circumstances, since the optimum speed for comfort
would be much higher. Moreover, softly sprung modem cars
would have dug troughs in the road to a depth of at least
80 mm. The vintage sports car would be bumping on its
suspension stops, if the driver attempted to get 'into the
groove'. Naturally if only cars with stiff suspensions ever
used this road, the problem would not arise.
2.5.1 Body resonance

In the previous example, we considered the case of resonance
of the suspension under the action of the unsprung mass.
Resonance of the suspension under the sprung mass can occur
and with no damping or inadequate damping this can be very
disconcerting. Let us consider the case of a car with an
undamped suspension system with a mean natural frequency
of 1.4 Hz which encounters a series of ripples with a
wavelength of 10 m.
At a velocity of 10 x 1.4 = 14 m/s = 50 k.p.h. (31 m.p.h.)
the forcing frequency induced by the regular road ripples
would exactly coincide with the natural suspension frequency
~8r-----~------'-----7r-----'
°O~~--~/~·O~--~2~·O~--~3~~~----47~
Frwquenc!I Ratio. ~I.fn.
Fig.2.6 Damping characteristics: note resonance whenfJfn is unity.
34 Springs
under the action of the sprung mass. Fig. 2.6 shows what
happens on a typical undamped suspension system when this
occurs. The undamped curve is indicated by the number 1.
Naturally on a washboard road the driver would accelerate
through the critical speed that caused body resonance. The
profile of a typical road surface, however, presents a wide
gamut of forCing frequencies and at any chosen cruising speed
a car with undamped springing would quite frequently strike
stretches of road that induced resonance.
Optimum damping is usually a compromise, a compromise
that is often influenced by the tester's individual preference.
Competition vehicles are sometimes fitted with adjustable
dampers and are set up to meet the requirements of the
particular driver. The contribution of the damper to
suspension design is of primary importance and will be
discussed fully in Chapter 7
2.6 The spring in the tyre
We have already commented on the effectiveness of the

pneumatic tyre in isolating us from minor road surface
irregularities. Even a road made of flat-topped paving stones,
so popular in the larger towns and cities before the First
World War, can be traversed in comfort with modem
low-pressure tyres. At a speed of 50 k.p.h. (31 m.p.h.) stone
setts of 6-in width would impart a vibration of about 90 Hz.
If the setts are well laid with no more than 10 rom (0.4 in)
variation in height between neighbouring setts, the flexing of
the contact patch and the side walls will almost completely
absorb the vibrations.
The complex construction of the modem tyre makes the
application of mathematics to the flexural behaviour of tyres
over a range of frequencies a daunting challenge. Even so,
better men that the present writer have tackled the problem
and Drs Overton, Mills and Ashley carried out a series of
experiments at Birmingham University [1] in which they
measured the dynamic behaviour of both cross-ply and radial
tyres (non-rolling), successfully constructing a mathematical
model that gives an accurate prediction of the experimental
results.
The spring in the tyre 35
Fig. 2.7 Mathematical model used in Birmingham University

experiments.
Distortion of the tyre carcase under cornering, acceleration

or braking forces, could of course modify their basic concepts,
but the major factors neglected by the fact that the
measurements were made on non-rolling tyres are: (a)
centrifugal expansion of the tyre carcase, and (b) the mobile
nature of the footprint compression. The first factor would
only make their predictions inaccurate at high speeds, but the
second factor is a valid source of error since the continuous
process of establishing a new footprint probably increases the
effective tyre stiffness. It does not mean, though, that the
critical vibration modes measured in these experiments are far
from the truth.
The simple suspension model chosen by Overton, Mills
and Ashley is shown in Fig. 2.7:
V 3 = velocity imparted to sprung mass;
F 3 = force imparted to sprung mass;
K3 = spring rate;
m = unsprung mass;
V 2 = velocity of unsprung mass;
F 2 = force given to unsprung mass (at wheel hub).
36 Springs
Fig. 2.8 Mathematical model for complete suspension system.
The 'springiness' of the tyre is represented by two springs.

The first Kl is undamped and can be regarded as the simple
compression of the air in the tyre. The second K2 represents
the characteristic behaviour of the tyre carcase. Modem
synthetic tyres exhibit a high degree of hysteresis. Expressed
simply, this means that the material is not perfectly elastic
and for every unit of kinetic energy absorbed by the rubber a
proportion is converted into heat energy.
It is unfortunate that many synthetic rubber mixes that grip
the road so well (high-p compounds) are also high-hysteresis
compounds. When taken to extremes the use of
high-hysteresis compounds creates problems for racing team
managers, since such tyres give excellent grip in the wet but
overheat in the dry. Modem road tyres are a reasonable
compromise, in that they give a good grip in the wet but do
not overheat in the dry. The degree of hysteresis is still
enough to give a fair measure of damping, and this damping
action increases with increase of frequency. Hence the
1600-
351b/in 2
1200
r::
S
..0
T
~ 800
z
It
fii
400
0+----.-----,----.----,----,----,----,
o 100 200 300 400 500 600 700
LOAD-Ibf
Fig. 2.9 Variation of stiffness with load and pressure for 5.60 x 13
cross-ply tyre.
1200
1000
"
~BOO
,i ///
T
~ 600
....
z
... /
~ 400
200",,//
00 5 Ib I~ 2'0 215 3'0 ~5

PRESSURE -lb/in 2
Fig. 2.10 Variation of stiffness with pressure for 165 x 13 radial

tyre.
38 Springs
1'0
0·1
k 3 =840
r-- ----...,
I m=l·n I
I c=305 I 0-0 Experimental.
- - - - Theoretical.
I ~100 k= 10 400 I
IL _ TY':!..T0de~ _ _
I
..J Pressure 15 Ib/in 2 •
Preload 500 Ibf.
System Excitation 27·9 Ibf.
0·01-j---.----,--.,--..,..-,----,--r_~_r___r---,r-_.__-T-_r__.
o 8 16 24 32 40 48 56
FREQUENCY -Hz
Fig. 2.11 Motion transmissibility across 5.60 x 13 cross-ply tyre.
addition of parameter q" frequency, to the proposed

viscoelastic model used in the Birmingham University study.
The Overton, Mills and Ashley model omits the suspension
damper, since this does not enter into the resonance
characteristics of the tyre. The complete suspension system
(one wheel only) is represented more accurately in Fig. 2.8.
The damping coefficient C3 is expressed as N/ms. This
1·0
0'1
~
0--0 Experimental
*]=840 - - - - Theoretical.
,---------,
I m-m I Pressure 15 Ib/in2 •
I tC-~15 I Preload 500 Ibf.
Excitation 27'9 Ibf.
I ~IOO *2=14200 I
I I
L _:!t.r!.!"~e~ ___ J
System
0'01 I I I I I I
o 8 10 24 ~2 40 48 50
FREQUENCY -Hz
Fig. 2.12 Motion transmissibility across 163 x 13 radial tyre.
assumes that the damping coefficient increases directly with

frequency.
Figs. 2.9 and 2.10 show typical measurements of tyre
stiffness, or spring rate, plotted against load for a 5.60 x 13
cross-ply and a 165 x 13 radial tyre. The cross-ply shows a
complex variation in stiffness. The radial tyre, however,
shows a proportional variation with pressure. This is an early
40 Springs
design of a steel-braced tyre with very flexible side walls.
More complex modern radials using mixed textile bracing
materials will probably show some variation with load.
2.6.1 Motion transmissibility

Motion transmissibility is defined by Overton, Mills and
Ashley [1] as the ratio V.jV 1, i.e. the ratio of wheel hub
velocity to footprint velocity. Figs. 2.11 and 2.12 show both
experimental and theoretical values for the two tyres at a tyre
pressure of 151bf/in2 (1.03 bar).
These results show that a tyre is a very effective
vibration-absorber at frequencies above 40 Hz. Road impulses
from a very low frequency up to about 20 Hz are transmitted
unimpaired; in fact, at some resonance frequency about 16 Hz
the velocity V 1 at the footprint becomes V 2 = 4V1 at the
wheel hub. This resonance frequency varies with tyre
diplensions, inflation pressure and internal construction, but
falls inside a range of 12-22 Hz.
At a speed of 50 k.p.h. (31 m.p.h.) a frequency of 16 Hz
would be given by a sinusoidal wave formation in the road
surface of 0.9 m (3 ft). This could be the cause of the
'boulevard bounce' experienced in the USA, where the
popular fat tyres bounce so readily at typical city traffic
speeds.
References
[1] Overton, J. A., Mills, B. and Ashley, C. (1969-70), Proc. [nst.

Mech. Engrs, 184, part 2a.
[2] Carlson, H. (1973), Spring Designer's Handbook (Vol I, Mechanical
Engineering), Dekker, New York.
3
Suspension Principles
3.1 Coupled suspension!
In Chapter 2 we concentrated on the behaviour of the front

end of a car under vertical accelerations produced by
irregularities in the road surface. For simplicity, we
considered the front end of the car as a discrete mass
supported on a main spring and an auxiliary spring (the tyre).
If we omit such rare examples as the Panther 6, which has
four steered wheels at the front, we can define an automobile
within the remit of this book as a single mass supported on
four wheels, and it is the behaviour of this integrated sprung
mass on its four suspension systems that is our concern in
this chapter.
In practice the most complex patterns of behaviour can
occur and at any moment in time all four wheels can be
moving up or down at differing frequencies, through different
amplitudes and with phase differences between frequencies.
With such formidable behaviour patterns it is not surprising
that many car manufacturers with extensive test and
development facilities still fail to strike the right balance by
the time the car goes into production and find it necessary to
modify the suspension design at a later date.
The large manufacturers have accumulated a vast pool of
data on suspension design and are able to construct an
accurate mathematical model before they design the
suspension system for a new product. Even so, the
suspension engineer can not achieve perfection using
conventional suspension techniques. He can decide to
sacrifice a measure of comfort to improve cornering, or he can
decide that straight-ahead passenger comfort is his major
priority as in a vehicle designed to convey expensive
cut-glassware or a president, trusting that the vehicle will be
driven round all bends with respectful decorum.
42 Suspension principles
Cen troid
Fig. 3.1 Simple coupled suspensions.
3.2 Bouncing and pitching
Side-to-side suspension interactions will be neglected in this

chapter. They are of course a major factor in cornering
behaviour, but the phenomena of bouncing and pitching are
experienced in the main when travelling in a straight line. In
this chapter the pogostick of the last chapter has been
replaced by two pogo sticks connected by a rigid bar as
shown in Fig. 3.1.
Pure bounce will occur if the front and rear sprung masses
are equal, the front and rear springs have identical
frequencies and identical rates and are in phase. We could
design a car to have a 50/50 weight distribution and identical
springs all round, but the phasing of the ripples in the road
surface are beyond our control. It is inevitable that a degree
of rocking or seesawing, or pitching as it is known in this
context, must often occur. The inertia of the sprung mass will
resist this pitching movement and the disposition of the
major components, i.e. the engine, transmission, fuel tank,
passengers, etc., that make up the sprung mass contribute to
the resistance exerted by the sprung mass in opposition to the
Fig. 3.2 Low polar moment of inertia.

Bouncing and pitching 43
6
1==:t;:::====::J~~~
Fig. 3.3 High polar moment of inertia.
pitching moment. If the major masses of the body

components tend to be near the centroid, as in Fig. 3.2, the
sprung mass offers less resistance to pitching than a body of
the same total mass that has the more massive components
situated at a greater distance from the centroid, as in Fig. 3.3.
Expressed technically, we say that the second case has a
larger polar moment of inertia. A moment of inertia is a moment
of the second order, i.e. a moment in which each small unit of
mass om is multiplied by the square of its distance from a
chosen axis:
I = ~ r2 om
= f r2 dm. (3.1)
The axis, in this case, passes through the centroid. In this

book we are interested in two polar moments of inertia. In a
study of pitching the axis is horizontal (see Fig. 3.4). When
we study the case of a car making rapid changes in direction,
i.e. a racing car passing through a chicane, the axis of the
relevant polar moment of inertia will be vertical.
K =Radius of"gyration
Fig. 3.4 Radius of gyration.

If the total sprung mass in Fig. 3.4 is M and we choose a
value of K such that
MK2 = f r2dm (3.2)
the quantity K is called the radius of gyration of the mass about

the axis. From Chapter 2 the period in bounce is given by
(3.3)
The period in pitching is influenced by the relative values of K

and the distances 11 and 12 between the wheel centres and the
centroid:
T' = 2~J(~ X 11~21j. (3.4)
If 11 = 12 = K and the front and rear spring rates K are equal,

T' = T.
Although many modern cars have the centroid well
forward, the value of K 2/(l1 x IJ is often close to unity. This is
more by chance than by deliberate design.
3.3 Suspension theoryi
When a rigid body, such as a car body, is mounted on

springs at both ends, and the front springs are subjected to
vertical forces through the front wheels, oscillations are
produced in both front and rear springs. A similar action
occurs when the vertical forces are applied to the rear wheels.
One of the earliest mathematical treatments of this reaction
between front and rear suspensions was presented by
Professor J. J. Guest [1].
For simplification, lateral forces are not considered.
Moreover, it is also assumed that both front wheels are
subjected to identical vertical forces simultaneously. The same
conditions of identical forces in phase also apply to the rear
springs. The mathematical model is, therefore, identical to our
example of two pogo sticks connected by a rigid bar, as
shown in Fig. 3.1. Professor Guest's model is shown at the
45
y--:!"""---
x
z
--~------~----------~B~-----Y
Fig. 3.5
top of Fig. 3.5. If a vertical force is applied at the front end of

the beam, the line XY which represents the centre line of the
body, will adopt an inclined position X'y', rotating about
point B. If a vertical force is applied at the rear, then rotation
will be about point A. Any pair of points that possesses this
relationship can be shown to satisfy the equation pq = ab. The
dimensions a and b are established by the relative rates of the
front and rear springs k j and k21 since kja = k2b. The spring
centre C is, therefore, the point of balance and a vertical force
applied at C will lift the beam with no tendency to tilt.
Points A and B are called elastically conjugate points by
Professor Guest. Since any value of p can be chosen, the
corresponding value of q is established by the relationship
pq = abo Therefore, there is an infinite number of such pairs of
points. A geometric construction to help determine the pairs
of points is given in the lower portion of Fig. 3.5. The
hyperbola pq = c 2 is drawn relative to the axes XCY. For any
chosen value of P, the point A is projected vertically
downwards to meet the 45-degree line CZ at A. Projecting
hOrizontally then gives the corresponding value of B 1 on the
hyperbola line. This can then be projected vertically upwards
to give the other elastically conjugate point B.
It will be seen that as A approaches C, point B moves to
the right and will move to infinity when A and C coincide.
Similarly, as B tends to C, A will also tend to infinity.
The next stage of the Guest construction is to establish the
dynamically conjugate points of the sprung mass. The moment
of inertia of the sprung mass about the centroid is MK2. Fig.
3.6 shows how the sprung massM for whichI g =MK2 can be
represented by discrete masses m 1 and m 2 at distances rand s
from the centroid G. To satisfy the equation rs = K2, the
following conditions must be met:
m 1 +m 2 =M
m 1r = m.p
m 1r2 + m.p2 = MK2.
As in the case of the elastically conjugate points, the
quantity r is chosen arbitrarily and the corresponding value
Mass M
G 1 = MK2
9
r s
m1 m2
X1
y ~--~~--~------------+----------------y
E
'-1--- 5 -----1
Fig. 3.6
Suspension theory 47
Y--~-+~~~~~--~~J--------------------~Y
Fig. 3.7
of s is found from the relationship rs = K2. Again there is an

infinite number of pairs of dynamically conjugate points. The
geometric construction is shown in Fig. 3.6.
The final stage is to combine the two systems. This will
give us the double conjugate points that will satisfy both
equations. This is shown geometrically in Fig. 3.7. Points H
and J, projected from Nand L (L being the point where the
two hyperbolas cross) are points that are elastically and
dynamically conjugate. These points are called by Professor
Guest the double conjugate points; there is only one pair.
3.3.1 Special cases'

When G and C coincide, but c2 is not equal to K2, the point of
intersection of the two hyperbolas moves to infinity. This can
be seen from Fig. 3.7. Point J is, therefore, also at infinity and
point H coincides with G and C.
The other special case would occur when G and C coincide
and c2 = K2. In this case the two hyperbolas also coincide and
the double conjugate points are virtually indeterminate. It
would be an interesting experiment for the inquiring student
to observe how such a model behaves in the laboratory. (The
writer admits he has always ignored this special case!)
To calculate the double conjugate points Hand J, we have
the following information (see Fig. 3.8):
pq=c 2 =ab. (3.5)
rs = K2. (3.6)
P = r + x. (3.7)
q = s - x. (3.8)
For any given case, we know the suspension rates kl and k2
and W, the wheelbase, which equals a + b. From this we
know:
a= __ k2_
kl + k2 .
b=_k_ 1 _
kl + k2 .
K2 is a quantity that can be estimated by the student or can
be measured accurately on special apparatus available to
manufacturers or at certain research establishments such as
the Cranfield Institute of Technology.
G can also be estimated in the drawing office (using such
computer facilities as may be available) or, if the vehicle
exists, by direct measurement of the axle weights on a
weigh bridge .
From the positions of C and G we know the value of x.
Substituting for p and q in Equation (3.5), Equations (3.7) and
(3.8) give us:
(r + x)(s - x) = c2 (3.9)
and substituting for s from Equation (3.6), we have:
(r + X)(~2 - x) = c 2
Hence:
Suspension theory 49
This gives two values for r. The negative root is, of course,
imaginary. There are also two values of s, a positive and a
negative. s can be found from Equation (3.6) by substituting
the value of r. It will be found that the negative value of s
equals the positive value of r and vice versa. The two double
conjugate points Hand J are now known.
3.3.2 Properties of double conjugate points

The essential property of double conjugate points is that a
force applied at one of them will produce no motion at the other.
These independent motions are not pure bounce, however.
Each end of the beam (or car body) makes an angular
oscillation about its own conjugate centre, i.e. the front end
about J and the rear about H. Unfortunately, if we try to
approach pure bounce at the front end of the car by
increasing the value of s, we increase the angularity of the
motion at the rear by a corresponding decrease in the value of
r. When road disturbances produce pitch, this additional
angularity only adds to the discomfort of the rear passengers.
Despite the help from headrests and reclining seats, the
human frame is much less disturbed by bounce in moderation
than by pitching. The frequency of the pitching is, of course,
critical. A rocking chair can be soothing, but only when the
frequency is low. A rapid rocking motion, i.e. of 2 Hz or
more, tends to put a strain on the neck muscles. A ride in a
short-wheelbase Jeep or Land Rover will illustrate this defect.
With conventional springing, the designer soon finds his
options very limited. An analysis of several modem
suspensions shows that many suspension engineers have
chosen double conjugate points that coincide with the wheel
centres. In this way bounce at either end of the car does not
produce any bounce movement at the other. Spring rates at
front and rear are usually chosen to give bounce frequencies
that differ by at least 10%. Since the position of the centroid
changes appreciably from the 'one-up' to the 'four-up' load, it
is impossible to make the double conjugate points fall
precisely over the wheel centres for every possible loading.
3.3.3 Spring frequencies about double conjugate points

When the front end of ,the car shown as a simple mathematical
model in Fig. 3.8 pivots about its conjugate point J, the
Fig. 3.8 Double conjugate points: a simple mathematical model.
rotation is resisted by the moment of inertia of the sprung

mass about the same centre. The moment of inertia about
G = MK2. The moment of inertia about J = I J = M(K2 + S2).
The restoring couple is supplied by the springs at front and
rear. The front springs operate on an arm, at = a - x + s. The
rear springs have an effective arm, ar = S - b - x.
The periodic time can be calculated from these two
quantities.
T = 27tJ( Moment of inertia about J )

J Restoring couple a bout J for unit angular deflection
(3.11)
(3.12)
Similarly, the periodic time about H is given by:
(3.13)
Example 51
Periodic time in pitch
T
p
=
2
xJ(Restoring couple
Moment of inertia about G \
a bout for unit angular deflection}
G
(3.14)
(3.15)
The corresponding frequencies are the reciprocals of these

times.
3.4 Example
3.4.1 Ford Fiesta S

This front wheel drive saloon is an excellent example of the
small popular European family car. The ride is very good for
such a small vehicle. The suspension is first analysed for the
case where no passengers are carried:
Data: one-up
Total massM t = 800 kg
Sprung massMs = 727 kg
Wheelbase W = 2.286 m
Front/rear weight distribution = 63/37
Front suspension rate kl = 21. 7 kN/m
.Rear suspension rate k2 = 25.0 kN/m
K2
Ql x 12) (estimated) = 1.0
25
a = (21.7 + 25) x 2.286 = 1.224 m
b 21.7
(21.7 + 25) x 2.286 = 1.062 m
11 = a - x = 0.37 x 2.286 = 0.846 m
12 = b + x = 0.63 x 2.286 = 1.440 m
K2 = 1.0 x 0.846 x 1.44 = 1.218
x = a-II or 12 - b = 0.378 m
c2 = a x b = 1.300.
From this infonnation, using Equations (3.6), (3.10), (3.12),
(3.13) and (3.15), we obtain the following:
Double conjugate point distance r = 0.846 m
Double conjugate point distance s = 1.440 m
Front suspension frequency in bounce F f = 1.55 Hz
Rear suspension frequency in bounce F r = 2.17 Hz
Pitching frequency F p = 1.96 Hz.
It is noted that the Fiesta suspension engineer has chosen

double conjugate points that coincide exactly with the
wheelbase, i.e.
r = 11.
s = 12 •
This, of course, only applies to the one-up condition. As
discussed earlier, when the double conjugate points are
placed exactly at the wheel centres, front end bounce
produces no bounce oscillations at the rear and vice versa.
Data: four-up
Sprung massMs = 945 kg
Wheelbase W = 2.286 m
Front/rear weight distribution = 54/46
Front suspensionk 1 = 21.7 kN/m
Rear suspension rate k2 = 25.0 kN/m
K2
(11 X 12) (estimated) = 1.05
a = 1.224 m
b = 1.062 m
11 = a - X = 0.46 X 2.286 = 1.052 m
12 = b + X = 0.54 X 2.286 = 1.234 m
K2 = 1.05 X 1.052 X 1.234 = 1.363
X = a-II or 12 + b = 0.172 m
c2 = a X b = 1.3.
From this data we obtain the following:
Double conjugate point distance r = 1.269 m
Double conjugate point distance s = 1.074 m
Example 53
Front bounce frequency F f = 1.45 Hz
Rear bounce frequency F r = 1.69 Hz
Pitching frequency F p = 1.56 Hz.
This shows a general softening of the suspension.
For a small car a rear bounce frequency of about 1. 7 Hz is
very satisfactory. The only criticism we can make is that the
rear end frequency in bounce is very close to that in pitch.
The student could attempt a few experimental designs to see
if he can improve on the design of the Ford Motor Co. He
must always remember to work inside the physical limitations
imposed by the overall body design. For example, the use of
lower spring rates increases the total wheel travel. The
intrusion of the upper spring mountings into the engine bay
at the front, and the rear seating at the rear, could create
acrimony between the suspension engineer and the body
engineer.
Reference
[1] Guest, J. J. (1925-6), 'The main free vibrations of an autocar',

Proc. Inst. Auto. Engrs (London), vol. 20, no. 505.
4
Suspension Geometry
4.1 Front wheel orientation
Over the years automobile engineers have theorized and

experimented with camber angles, castor angles, kingpin
inclination and offset and front wheel toe-in and still there is
no consensus of opinion or general rules that can guide the
student. We can only examine the steering layout and
suspension geometry of such modern cars that are known to
handle well and hope to glean useful pointers. Fig. 4.1 shows
the two wheel angles, camber and toe-in and the two kingpin
(or swivel) angles. In the left-hand view the car is viewed
from the front, in the right-hand view from the side and in
the lower from above.
4.1.1 Toe-in
This is a very small angle made by each front wheel plane
and the longitudinal axis of the car. In Fig. 4.1 the angle is
much exaggerated for the sake of clarity. The amount of
toe-in is usually measured as a difference in the distance
between right and left wheel rims at front and rear, both
measurements being made at hub level. A typical toe-in
would be as little as 3 mm. Without toe-in the inevitable
compliance (or 'free play' as a typical mechanic would call it)
in the several ball joints used in the steering linkages, could
easily lead to the phenomenon known as 'shimmy', where
the wheels flutter in and out within the limits set by the .~otal
compliance. The action of toe-in is simply to keep all the
steering linkages under tension. Toe-out would also take up
the slack and give stable straight-line running. The track rod
and other links would be under compression in this case.
This is undesirable with relatively long tubular links. If the
manufacturer's specified toe-in is exceeded, excessive tyre
wear will occur. The optional toe-in is usually established
experimentally.
Comber
Thearetica I center
of rotation of cambered
wheel
I --- ---- ---
(Camber angle is neglected

in this view for clarity)
Fig. 4.1 Wheel orientation with the ground.
56 Suspension geometry
4.1.2 Camber
The camber angle is the angle made by the plane of the
wheel to the perpendicular as viewed from the front. It is
usually termed positive camber when the top of the wheel leans
outwards away from the car centre line. With modem car
tyres, camber angles cannot be large if an adequate footprint
area is to be maintained. A small amount of positive camber
at the front is traditional, since with no camber at the rear,
using the traditional live rear axle, positive camber reduces
\
I
\ \ I
\
\ \
\
!s:&
B
Understeered
c
Neutro I
A
Oversteered
Fig. 4.2 Effect of slip angles on directional stability. Vehicles A, B
and C are subjected to identical side forces. The right-hand car A
has greater slip angles at the rear than the front. This creates a
centrifugal force C acting in the same direction (apprOximately) as
the side force F. This calls for rapid steering correction by the driver.
The left-hand car C has greater slip angles at the front. In this case
the centrifugal force opposes the initial side force. This is a stable
situation. Neutral steer, as shown by car B, is given when slip
angles at front and rear are identical. In this case the car suffers a
sideways displacement, but does not steer to right or left.
Front wheel orientation 57
cornering power at the front relative to the rear. A car with
greater slip angles at the front than at the rear under the
action of a transient side force will be an understeering car. A
small degree of understeer is essential for good straight-line
stability. The reverse of understeer is called crversteer (see Fig.
4.2). This is discussed in more detail in Chapter 6.
Camber angle can change with suspension movement and
this is a critical design parameter when we come to layout
the geometry of the suspension linkages. Camber angle is
usually specified by the manufacturer in the unladen static
condition. Static angles seldom exceed two degrees.
4.1.3 Kingpin inclination (or swivel angle)

The term kingpin survives from the days before independent
suspension when the stub axle on which the wheel rotated
Fig. 4.3 Details of kingpin and stub axle on beam front axle.
9°
Cas~or ° -----.I SwiveL-
angle --\ 7r--
\ -
- - - I~g lel
,0 --; '---,
Camber I :
I
angle I '~ il
· :
,I ,
i. (
11 3tl;i~ii6r--
Fig. 4.4 Modern steering layout, using upper and lower swivelling joints to replace kingpin.
Fro_nt wheel orientation 5(}
14---+---- e': e( cos a-)
King pin
inclination
Theoretical plane
of wheel rotation
Fig. 4.5 Effect of kingpin inclination on steering. When wheel is

steered around inclined kingpin, the axle is lifted, creating an
unstable condition. This provides a part of the self-aligning torque.
was arranged to swivel around a hardened steel pin mounted

in the end of the beam axle (see Fig. 4.3). With the
introduction of independent front suspension the kingpin
survived for a time, but was eventually replaced by a system
in which the hub carrier swivels about an upper and lower
ball joint, as shown in Fig. 4.4.
The kingpin or swivel angle is chosen to give the desired
offset, as shown in Fig. 4.1. The extent of this offset
determines the amount of self-aligning torque exerted when
the steering wheel is turned. This is illustrated in Fig. 4.5.
Turning the wheel, raises the wheel hub relative to the
ground. A moment is thus created which tends to return the
wheel to the straight-ahead position. Some modern cars have
been given negative offset. This reduces steering 'feel' at the
steering wheel, but is introduced to give stability at speed in
the event of a tyre blowout or brake failure on one front
wheel.
4.1.4 Castor angle

Castor angle also creates self-aligning torque, since it places
the contact patch behind the swivelling axis (see Fig. 4.1).
Cars with a kingpin angle that gives a negative offset will,
therefore, need a larger castor angle to compensate.
4.1.5 The dynamics of self-aligning torque

Self-aligning torque defies the rules of simple geometry, since
the behaviour of the tyre footprint exerts a major influence as
the slip angle rises. For moderate values of cornering force
self-aligning torque increase with slip angle. At extreme slip
angles the self-aligning torque is reduced, as shown in Fig.
4.6. Only when cornering at racing speeds is this reversal
encountered. To a racing driver, this loss of 'feel' at the
steering wheel is an indication that he is now driving 'on the
---
limit'.
6- _
_ Sliding
- - Racing
driver ~
5
'"~ ,I)o~,~
4 \
3
I
Skilled
driver
2
jt
/
/'.1'0
.,,/ oI'~
Average
driver
-100 -80 -60 -40 -20 60 80 100
Self- aligning torque (N m)
Fig. 4.6 Relationship between cornering force and self-aligning
torque on a typical tyre.
Front wheel orientation 61
Steering arm
Track rod
Ei3 ,
- I
------ _---:4--- ,
- I
,£:+3
Fig_ 4.7 Ackermann steering layout_
4.1. 6 Ackermann steering layout

Ackermann arranged his steering layout to give a slightly
greater steering angle to the inner wheel than the outer. His
original layout is shown in Fig. 4.7. The steering arms are
inclined inwards so that their projected lines converge to meet
at the centre of the rear axle_ If one neglects the effects of slip'
)¥-r%
, \
/ - , ,
. .-/ . ,.......... ---------- I
,/ "..___- I
/ I,
'/ I
. ../
e/--?-' -----_. __ WIT

Ackermann
centre
01 --+ _ 1
I
.
,
rill'
,
.
Fig. 4.8 Ackermann steering centre (with zero slip angles)_

angle (as Ackermann obviously did), the car will have a
turning centre on any comer which is in line with the rear
axle centre line (see Fig. 4.8). When slip occurs at the four
contact patches the turning centre will move forward, the
extent of this movement depending upon the relative slip
angles at front and rear. Not all designers adopt the original
Ackermann layout. Some use an intersection point behind the
rear axle, others prefer parallel steering arms. There is some
logic in adopting a negative Ackermann angle, i.e. one where
the intersection point lies ahead of the front wheels. High
cornering forces create roll and demand greater slip angles
from the outer tyres than the inner. The original Ackermann
layout, designed to give a greater angle to the inner wheel, is
therefore based on a false premise. The dynamic behaviour of
a car when cornering will be analysed in Chapter 6.
4.2 Roll geometry
4.2.1 Roll resistance

When a body rolls under centrifugal force, this force acts at
the centroid of the sprung mass (see Fig. 4.9). The body
rotates about the roll centre and this most important parameter
r;i\
Roll
angle
~._F T,
G~I'
h '
" Mo'
Roll
centre
Fig. 4.9 Roll moment.
Roll geometry 63
is defined by the geometry of the suspension system, as will
be discussed later in this chapter. The roll moment which
must be resisted by the suspension springs (and the anti-roll
bars, when these are fitted) is equal to F x h, where F is the
centrifugal force and h is the distance between the sprung
mass centroid and the roll centre. For a given spring rate, the
roll resistance will be greater and the roll angle less if h is
reduced.
4.2.2 Roll centres

For a typical double wishbone suspension the position of the
roll centre can be determined as follows. For simplicity, only
one end of the car is considered at this stage. Alternatively,
we could say that the front and rear suspensions are
considered to be identical, the weight distribution 50/50 and
the axial centroid centre line horizontal. Under a side force F,
the body will tilt and the spring on the left will be
compressed to increase the vertical load R} (see Fig. 4.10). We
could produce the same effect by holding the body in a
horizontal position while rotating the wheel L about the
instantaneous suspension centre C L relative to the body. This
centre C L is given by the intersection of the projected centre
lines of the upper and lower wishbones. The wheel footprint
x
I
I--~
FIG Wheel R
rP==_=-1--:- .- - or -----
-==---
0 0~~ ~
_ -' _
-=-- - - - --:- --.

---::::::..:...~ -:-===-- I
I
Xl --- --
v --
Fig. 4.10 Determining the roll centre.
will move at right angles to line PCL to a new position Q. At
the same time the decrease in load R2 will cause the contact
point V of wheel R to move to position V. If we rotate the
body in an anti-clockwise direction until the wheel footprints
have returned to their original positions P and V, the only
point about which rotation can occur within the constraints
that wheel L must move about centre CL and wheel R move
about centre CR is seen to be point Mo. The position of Mo is
found by the intersection of lines drawn between CL and P
and C R and U. In some cases, notably the popular
MacPherson strut suspension, the roll centre is not at a
constant height above ground and moves slightly as the roll
increases.
4.2.3 Typical roll centres

The roll centre on a beam axle supported on semi-elliptic
springs is decided largely by the curvature of the springs and
the position of the shackles. In general it is situated
approximately halfway between the spring base and the
upper shackle pin, as shown in Fig. 4.11.
Three systems have the roll centre at ground level. The first
is a double-wishbone system with parallel links of equal
Fig. 4.11 Roll centre on beam axle with semi-elliptic springs .
• - - • - - - --11-1<>--1-
-- -- - -r-!-'~--t--~::t-<>I
. - - -- - - ...,....,....,~...,......,.....,.-.,rl......."....,~...,...-r-~~/
M"
Fig. 4.12 Roll centre with double-wishbone suspension of equal-
length parallel links.
Roll geometry 65
Mo
Fig. 4.13 Roll centre with vertical sliding pillars.
length (see Fig. 4.12). Since the instantaneous link centres are
at infinity, the roll centre must be at ground level. The second
system is the old Porsche trailing link system using
equal-length parallel links as used on th~ familiar VW Beetle.
This system is shown in Fig. 5.10 in the next chapter. The
third is the sliding pillar suspension with vertical pillars, as
shown schematically in Fig. 4.13. When the sliding pillar is
set at a swivel angle (see Fig. 5.9 in the next chapter), the roll
centre falls below ground centre. The position is defined by
extending the pillar centre line to intersect the ground plane,
then by drawing a perpendicular line to intersect the vertical
centre line of the car.
The de Dion suspension system separates the two functions
normally performed by a live rear axle. The final drive unit is
mounted on the chassis (or body) and is, therefore, part of
the sprung mass. Open halfshafts with universal joints at
both ends take the drive to the wheels. The de Dion tube
which connects the two wheels and maintains them in perfect
parallelism is always located laterally, either by a vertical
slide, as shown in Fig. 4.14, or by other effective means.
With effective lateral location, the body is constrained to roll
about the centre point of the de Dion tube.
The swing axle is now obsolescent. It was used by
Profes.sor Porsche for the rear suspension on many of his
Fig. 4.14 Roll centre with de Dion system.

Fig. 4.15 Roll centre with swing axle suspension.
designs, almost to the point of obsession. The system oecame

notorious for excessive roll-steer effects that militated against
safe handling. Fig. 4.15 shows that the roll centre is very
high. This, of course, tends to reduce the roll moment (see
Fig. 4.9) and it is theoretically possible to design a racing car
in which the centroid of the sprung mass coincides with
the roll centre. This, at first sight, is an ideal system.
Unfortunately the cornering force, acting at road level on the
footprint of the inner wheel, tends to twist the swing axle
and lift the inner end. This produces what is called a 'jacking
action'. Fig. 4.15 shows how this can happen at high
cornering forces. The Daimler-Benz Co. eventually produced
a low-pivot swing axle design. It was a complicated, but very
effective, design (see Fig. 4.17) and was eventually replaced
by the semi-trailing link design which is so popular today and
which will be described in Chapter 5.
4.2.4 MacPherson strut

This type of suspension - sometimes called a Chapman strut
- is well illustrated in Fig. 4.18 and the schematic drawing in
Fig. 4.19 shows how the roll centre is found. The hub carrier
slides up and down the strut but is attached rigidly to the
outer tubular member. The strut must be capable of a slight
angular movement at its upper attachment point, since the
lower link moves through an angle under bump and
rebound. Some compliance is, therefore, necessary in the
upper attachment. This is usually given by the provision of a
rubber mounting. Some designers also provide a ball thrust
bearing, others use a low-friction thrust washer. This permits
strut rotation as the wheels are steered.
Fig.4.17 Low pivot swing axle suspension, as used on Mercedes-Benz W196 racing car.
r-
Roll geometry 69
9'
SWIVEL
ANGLE
I'
CAMBER
ANGLE
Fig. 4.18 Chapman strut suspension on early Lotus sports car.
Fig. 4.19 Determining centre on MacPherson (Chapman) strut

suspension.
4.2.5 Roll axis

The roll centres at the front and rear are seldom at the same
height. Similarly, the static loads carried at the front and rear
are not often equal. Fig. 4.20 shows how the sprung mass
can be treated as two discrete masses, M f at the front and Mr
at the rear. The centroid axis is at an angle relative to the roll
FD o
M, ·- .---SentrO
- _ _ _ R OI(
. ~ . ox's
G-=--- ·-.....;...."O"'is
Fron l roll
id .
---.:... .
rOUi'>(j ·:-:-....... _ _ .
Pion c :::-..- ___
entre_/,:--"
- ._ .-
@
~ ,
M,
. ___ ~ l /
F
me 7~
~ centre
Reor roll
Fig. 4.20 Relationship of centroid axis to roll axis.
axis, this angle being determined by the heights of the front

and rear masses M f and Mr above their respective roll centres.
In the particular example the roll moment is obviously greater
at the rear. The suspension system must be designed to cope
with this imbalance.
4.3 Anti-roll bars
Anti-roll bars often supply a simple solution to the problem

of roll imbalance. It is a simple device, being a transverse bar,
cranked at both ends in the same direction and clamped to
some convenient point on one of the suspension links on
opposite sides. They can be fitted to front or rear suspensions
or to both.
It will be seen in Fig. 4.21 that body roll will cause one
cranked arm to move downwards, while the other moves up-
wards. The anti-roll bar is, therefore, a simple torsion bar. If
however both wheels at the same end of the car rise or fall in
unison, no torsion is applied to the bar. Under the action of
single wheel bounce, the main spring on this side works in
parallel with an effective additional spring which is the
combination of the anti-roll bar and the main spring on the
opposite side working in series. To clarify what this statement
means, let us first consider what happens when two springs
of different rates act together side by side, i.e. in parallel. For
Anti-roll bars' 71
Fig. 4.21 Anti-roll bar on Ford Granada/Consul.
springs in parallel the rates are additive:

k p =k 1 +k 2 (4.1)
where k p is a single spring of equivalent rate to the two
springs of rates k 1 and k 2 acting in parallel.
When two springs act in line, i.e. in series, the equivalent
spring is always of reduced rate. An equivalent spring of
stiffness k s to two springs k 1 and k 2 acting in series must
produce the same extension under the same load:
Therefore
(4.2)
4.3.1 Single wheel bounce (with anti-roll bar)
Applying the above to the case of single wheel bounce, we
can take the main spring rate at the wheel as k and the
anti-roll bar rate, also measured at the wheel, as k.. The
single wheel bounce rate ke is, therefore, given by:
ke = k + kka k . (4.3)
a+ k
A good example to illustrate how the anti-roll bar can
influence the single wheel bounce rate is the Porsche 928,
which is the subject of a design analysis in Chapter 12. The
Porsche 928 is provided with a strong anti-roll bar at the
front and a relatively weak one at the rear. The appropriate
data are as follows:
Front Rear
Wheel rate (kN/m) 18.63 22.55
Anti-roll bar rate (at wheeI)(kN/m) 83.4 10.2.
Single wheel rates in bounce
Front 83.4 x 18.63
18 . 63 + ----,-----,--~
=
83.4 + 18.63
= 33.86 kN/m
Rear 10.2 x 22.55

22 .55 + ~-=----::-:-=-=-
=
10.2 + 22.55
= 29.57 kN/m.
This demonstrates the penalty involved when relatively stiff
anti-roll bars are resorted to as a means of limiting roll angles.
4.4 Anti-dive and anti-squat
Without resort to anti-dive and anti-squat geometry in the

suspension layout, cars with soft suspension will dip at the
front and rise at the rear under heavy braking. During hard
acceleration the reverse effect will occur. It was not
uncommon more than twenty years ago to see American
Anti-dive and anti-squat 73
Fig. 4.22 Anti-dive system, using leading arms at the front and
trailing arms at the rear.
sedans perform something that resembled a curtsy when

stopping at the traffic lights. A source of embarrassment was
the locking of front bumpers under the rear bumpers of the
car immediately ahead. It was no easy task to separate the
locked cars. The provision of over-riders on bumpers
prevented this particular danger but the physical discomfort
still remained. Brake dive is felt by the passenger as a
pitching motion and this puts more strain on the neck
muscles than is experienced with simple deceleration in the
horizontal plane.
A simple anti-dive design could be given by the use of a
leading arm at the front and a trailing arm at the rear, as
shown in Fig. 4.22. Under braking action the backplates (or
callipers, when disk brakes are fitted) tend to rotate with the
wheels. This produces an upward reaction at the front of
Fig. 4.23 Angled double-wishbones to give effective leading arms

to front suspension.
pL 11-pi L
pL 11-piL
o
y
Fig. 4.24 Anti-brake dive geometry: (a) with outboard brakes;

(b) with inboard brakes.
the body and a downward at the rear. With the correct

proportions, this can be designed to give full anti-dive
reaction.
Fig. 4.23 shows how the upper and lower links of a
double-wishbone front suspension system can be angled to
give an effective leading arm length. Fig. 4.24 applies to such
a case. It is assumed that the front braking effort does not
vary from the designed front/rear bias. If the front/rear
braking effort is designed to give an effort ratio of 60/40, the
value of p in both upper and lower diagrams of Fig. 4.24 will
be 0.6.
The upper diagram applies to the case of a car with
outboard brakes. Lines are drawn from the front and rear
footprints to point O. This point is on the sprung mass
centroid axis, the position relative to the front and rear
wheels being determined by the braking ratio p/(l - p). For
Anti-dive and anti-squat 75
full anti-dive correction, the effective pivot points, X for the
front suspension and Y for the rear suspension, should lie on
the appropriate lines. With inboard brakes, since the brakes
are part of the sprung mass, the braking torque is applied
through the drive shafts. The suspension pivots in this case
must therefore lie on lines drawn between the wheel centres
and point 0, as shown in the lower diagram.
Anti-dive geometry is not without its problems. Since the
upper and lower wishbones pivot about non-parallel axes,
this results in a change in the front castor angle with bump
and rebound. Some designers find this castor angle variation
unacceptable and compromise by correcting only a percentage
of the brake dive. The Jaguar XJ-S, for example, is designed
to give only 50% anti-dive correction.
4.4.1 Anti-squat
Anti-squat is resistance to squatting down at the rear end, and
as we may see in its more spectacular forms on the drag
strips it can be resisted in the limit by the provision of rollers
at the rear extremities of the body. Fortunately for the
designer of more ordinary vehicles, the provision of 'leading
arm effect' at the front and 'trailing arm effect' at the rear
works well in the reverse direction. Some compromise is
necessary, since an anti-dive geometry carefully calculated to
match a particular front/rear brake distribution will seldom
give a perfect correction for anti-squat.
In Formula 1 racing the problems of anti-dive and
anti-squat became acute in 1979, when the majority of
Formula 1 cars adopted 'ground-effect' body designs that
pulled the car down on the road by the vacuum created on
the underside of large side pods. Sliding skirts were devised
to seal the body sides to prevent leakage of higher-pressure
air to destroy this vacuum. These skirts actually touched the
road surface and often made grooves in the tarmacadam
when the suspension exceeded the available movement of the
skirt in its slides.
The most successful designs of the 1979 racing season were
those that achieved the closest approach to a no-roll
suspension with negligible dive and squat. The latest type T4
version of the Ferrari 312 uses heavy Bowden cables running
from front to rear to give a coupled front/rear suspension
system designed to reduce dive and squat. As I write, the
outcome of the experiment is unresolved, but at least they
have won the Constructors' Championship for the year.
5
Conventional Systems
5.1 The beam axle
Less than fifty years ago the beam axle suspended on

semi-elliptic springs was the norm at both ends of the
automobile. There were a few exceptions. Sizaire-Naudin
used a sliding pillar independent suspension at the front as
early as 1906 and a few years later the Morgan three-wheeler
and the Lancia Lambda appeared with refined examples of
the basic design. Even so, these were regarded as rather
expensive eccentricities and did little to convince the bulk of
automobile manufacturers that the simple rigid front axle
mounted on leaf springs, and the much heavier rear axle
casing with similar springs, were not a final solution to the
problem of springing. All that was needed, or so it seemed,
was a measure of refinement and a gradual reduction in the
manufacturing costs.
Despite this air of self-satisfaction, many experienced
motorists knew that both front and rear axles could
sometimes develop their own form of 'shakes'. At the front it
was called shimmy, at the rear tramp. For good measure, some
front axles were capable of both phenomena.
5.1.1 Shimmy
The driver could feel the onset of shimmy when the steering
wheel began to vibrate rapidly from side to side. This,
precisely, was what was happening to the front wheels.
Accessory manufacturers made steering dampers designed to
reduce the amount of shimmy. In racing and sports cars the
designer provided very stiff springs to reduce the angular
deflection of the front axle, since this angular deflection of the
axle is the source of shimmy.
Any rotating mass, such as a wheel and tyre, exhibits
gyroscopic effects. Thus, if a spinning wheel is subjected to
78 Conventional systems
an angular displacement, a force is produced tending to move
the axis of rotation in a direction at right angles to the original
displacement. The reader can demonstrate this phenomenon
using an ordinary bicycle, preferably one with large-diameter
wheels. He should stand astride the bicycle and lift the front
wheel clear of the ground. Then holding the cycle in this
position with one hand only on the handle bars, spin the
front wheel as fast as possible in the normal forward
direction. Finally, with the wheel spinning fast, he should
grasp the handle bars with both hands and lean the cycle to
the right. The gyroscopic torque will be felt as a distinct
reaction at the handle bars as the wheel tries to tum to the
right. If the wheel is spun in the reverse direction, the
steering reaction will be to the left.
This serves to illustrate the gyroscopic torque that twitches
the front wheels to left or right when passing over a bump or
a pothole in the road (see Fig. 5.1). With inadequate damping
of the front suspension (a characteristic of early vehicles), the
inertia of the heavy front axle and the masses of the wheels at
each end induces an oscillation at the natural frequency of the
springs. As long as these oscillations continue, with one
wheel rising as the other falls, the gyroscopic torque will
produce the steering oscillations of shimmy.
5.1.2 Tramp
With a beam axle tramp can occur at front or rear. Even when
shimmy is eliminated a beam axle can still develop tramp,
which is a transverse rocking action at spring frequency that
can persist for many seconds. The most common cause of
oc.=Angu/ar def'/ection
of' axis of' wheel
Fig. 5.1
The live rear axle 79
Braking
torque·
Fig. 5.2 Axle twist under braking torque.
front axle tramp is induced by hard braking. Even the great

Henry Royce was shaken physically and mentally when he
tested the first Rolls-Royce to be fitted with front wheel
brakes. Brake tramp was so violent that the headlamp bulbs
failed as the filaments broke. For a time Henry Royce was in
the dark! Improved location of the front axle was the eventual
solution.
The conventional location of semi-elliptic springs at this
time was through the single shackle pin at the front.
Variation in spring length was accommodated by the
provision of swinging shackles at the rear. Under braking
torque the springs could 'windup', i.e. twist into a flattened
S-shape as shown in Fig. 5.2. Spring windup reduced the
castor angle. In extreme cases it could become negative,
resulting in a negative self-aligning torque.
I remember a bad case of brake tramp when a vintage car
was raced for the first time at a VSCC Silverstone meeting.
Hard braking causes such severe brake tramp that the mag-
nificent mahogany dashboard was cracked from side to side.
Some deSigners of vintage sports cars used torque rods to
prevent spring windup. Others found an easy solution in the
adoption of very stiff springs that gave a very hard ride.
5.2 The live rear axle
All automobile designers have now adopted independent

suspension for the front wheels. A beam axle at the rear is
usually called a live rear axle when it also acts as the housing
for the final drive unit and axle shafts. The tenn dead rear axle
has been adopted to describe the use of a beam axle between
the rear wheels on a front wheel drive car.
5.2.1 Disadvantages
The live rear axle is relatively heavy in comparison with an
independently suspended one. The outer casing is
constructed from malleable castings and/or steel pressings and
contains the final drive crown wheel and pinion, the
differential casing and its gears, the halfshafts and all the
necessary bearings to support the moving parts. The
customary drum brakes, brake shoes, backplate, wheel hubs,
wheels and tyres all contribute to a fonnidable unsprung
mass.
The unsprung mass is approximately twice that of a good
design of independent rear suspension. This in itself is a
serious disadvantage as was demonstrated in Chapter 2. A
car with a low ratio of unsprung-to-sprung mass (m/M) will
maintain tyre/road contact on a rough road much more
effectively than one with a high ratio. In Chapter 2 it was also
shown that a high m/M ratio is also associated with a hard
ride.
Apart from the low cost of the live rear axle, it does
represent a very robust design and with the help of modem
techniques of longitudinal and transverse location satisfactory
rear suspension can be given for a low-cost front engine/rear
drive family car. The word 'satisfactory' is used advisedly,
since the high standards established by a good design of IRS
cannot be matched.
5.2.2 Five-link system

Paradoxically this popular location system for the live rear
axle owes much to the experience gained in the development
of IRS systems intended as replacements for the live rear axle.
The five-link system uses two non-parallel unequal-length
trailing links per side to give longitudinal location of the axle
(see Fig. 5.3). Parallel links of equal length would give the
closest approach to vertical motion of the axle under bounce
Fig. 5.3 Five-link suspension on live rear axle.
and rebound, but the links are usually given a convergence

on an effective pivot centre, thus giving a measure of
anti-dive effect. The fifth link is required to give lateral
location. A Panhard rod, as shown in Fig. 5.3, does not give
perfect location but is a simple solution. The rod should be as
long as possible to constrain the axle to move in a 'flat arc'.
One end of the rod is attached to a bracket on the
body/frame, the other to a bracket on the axle casing on the
opposite side. Rubber bushes are used on all five links to
reduce the transmission of vibrations to the body.
There are alternative systems of lateral location. A popular
one giving perfect lateral location over a limited range, is that
Fig. 5.4 Watt's linkage location for de Dion suspension.

invented by James Watt. Fig. 5.4 is a simple diagram to
explain the principle of this linkage. With such excellent axle
location, one can replace the heavy semi-elliptic spring with
the lighter, less expensive, coil spring. This again is a
technical spinoff from the development work on independent
suspensions.
5.2.3 Four-link system

This system dispenses with the lateral location link .by
arranging the 'longitudinal' upper links at an angle relative to
the centre line of the car. Such a layout is not likely to give
such accurate lateral location as that provided by a Watt's
linkage.
5.2.4 Two-link plus A-bracket system

This is a variation on the above theme. There are two lower
links (as in the above systems), but the upper links are
integrated into a single A-bracket pivoted at two points on
the body and one above the centre point of the axle. The
Renault, shown in Fig. 5.5, is a good example of this system
applied to a dead rear axle.
Fig. 5.5 Two-link system with A-bracket.

5.2.5 Torque tube
Since the early designs of an open propeller shaft were so
unreliable, it was common to enclose the drive in a casing or
torque tube and let this and the whole rear axle casing pivot
about a spherical bearing at the rear end of the gearbox. The
complete T-shaped assembly was thus given an accurate
location, even though the unsprung mass was increased
considerably by the long torque tube.
Hotchkiss was a pioneer in the use of an open propeller
shaft, using the semi-elliptic springs to locate the rear axle.
Surprisingly he also pioneered the use of a short torque tube
in conjunction with a short open propeller shaft in the 1922
type AR Hotchkiss. In this large luxury car a cruciform
cross-member is provided approximately halfway between
the gearbox and the rear axle. The spherical joint of the
torque tube is mounted at this station. This not only reduces
the mass of the torque tube, but removes the whirling
dangers associated with the use of very long open propeller
shafts. There is, therefore, much to recommend the layout
Torque
~ube
Telescopic damper
plus auxiliary spring
Fig. 5.6 The Rover 3500 live rear axle with short torque tube.
chosen for the Rover 3500 as shown in Fig. 5.6. The coil
springs are placed immediately above the axle tubes. Forward
of each tube are combination spring/damper units. These are
located on forward-facing brackets at points on the lines
drawn between the torque tube universal joint and the wheel
centres. In this position the spring/damper units can control
body roll as well as bump and rebound. Lateral location is
given by a Watt's linkage.
5.3 The de Dion axle
The invention of the de Dion axle is usually attributed to

Comte Albert de Dion, but it was actually conceived by two
engineers, Trepardoux and Bouton, who worked for him. A
well-engineered design that serves to illustrate the principle is
that used on the 1938 type W154 Mercedes-Benz Grand Prix
car (see Fig. 5.7). The wheel hub carriers are coupled
together by a cranked de Dion tube, which in this design can
pivot at its centre point and is constrained to rise and fall in a
vertical slide. The sliding block is pivoted at the front of the
tube, while the slide is incorporated in the final drive casting.
Unlike a live rear axle, the final drive unit is carried on the
chassis and is thus part of the sprung mass. The drive
to the wheels is by universally jointed halfshafts. A single
ball-Jointed radius arm is attached to the top of each wheel
hub carrier. To permit single wheel bump and rebound,
provision was made for relative movement between the right-
and left-hand halves of the de Dion tube. The provision of
twin parallel, equal-length trailing links would have removed
the need for this complication (see Fig. 5.8).
The de Dion axle retains the one advantage of the beam
axle, in that the wheels are maintained parallel at all times.
When cornering on a good surface, both wheels remain
vertical. This was advantageous in 1938. In the 1980s, with
more and more cars using low-profile tyres, it is still an
excellent feature. The other advantage is that the unsprung
mass is much reduced in comparison with that of a live rear
axle. As will be shown later, it is difficult to achieve as low
an mjM ratio as that given by the lightest designs of
independent rear suspension.
Independent front suspension 85
Fig.5.7 The 1938 Mercedes~Benz type W154 de Dian axle.
Aston Martin have used a de Dion rear axle on their cars

for years with excellent results. As will be seen from Fig. 5.8,
the unsprung weight has been kept to a minimum by the use
of inboard disk brakes. The provision of parallel trailing arms
on each side also makes it possible to use a one-piece de Dion
tube. Transverse location is given by a Watt's linkage.
The drive shafts are fitted with universal joints at both
ends. Either the inner or the outer joints must permit some
axial movement. On the Aston Martin the variation in drive
shaft length under bump and rebound is accommodated by
the provision of sliding splines inside the inner universal
joints.
5.4 Independent front suspension
5.4.1 Sliding pillar system

We have already said enough about the disadvantages of a
live front axle to show how necessary it was to replace it.
Sliding pillar IFS was first used in 1906 and, if we neglect the
rather primitive designs that appeared on early Morgan
three-wheelers, it appears that the first successful design
appeared on the Lancia Lambda, first shown at the Paris
GO
u5
....cb
""
Swivel
angle
20
camber
angle
\ \. . -. - ~
.-" --~ --
: 1
...I
I· I
I
.,-v--r-- ~ =='.-=.".-=-
I.
i *I
, .'
--.-_. /"' " Roll
centre
Fig. 5.9 Morgan sliding pillar IFS"

Salon in 1922. Even by modern standards, the road-holding
of this vintage treasure is very good. On the Lambda each
wheel hub carrier moves up and down on a sliding pillar.
Since the pillars are integral with a transverse tubular frame
that is part of the chassis, the unsprung weight given by this
system is probably the lightest ever devised. The Lancia
system was very sophisticated. The coil springs were
enclosed and tubular hydraulic dampers were incorporated.
(Unfortunately, I am not able to find a suitable illustration of
this excellent example.)
The only surviving example of this sliding pillar system is
seen on the Morgan sports car. From the earliest designs
dating back to 1910, the Morgan Motor Co. has gradually
developed the well-engineered layout shown in Fig. 5.9.
Single wheel movements in bump or rebound are vertical,
with a small reduction in track, as shown by the broken line
in the figure as the wheel rises. When the car rolls, however,
both wheels adopt new camber angles directly related to the
roll angle. In the example shown with static camber angles of
2°, a role of 3° when cornering hard would give a camber of
50 on the outer wheel and lOon the inner. The reverse
would be more acceptable with modern low-profile tyres. To
combat this disadvantage, relatively stiff springing is
necessary on the Morgan.
5.4.2 Trailing link suspension

Professor Ferdinand Porsche made this system memorable
when he adopted it for the Volkswagen Beetle. Fig. 5.10
shows an early design of Porsche front suspension as used on
the type 356B Porsche. With the tyres of the period (area
1960) with aspect ratios no less than 85% the contact patch
was well maintained and suffered little loss of area up to roll
angles of 5° or 6°. This of course is only one aspect, since a
contact patch becomes less effective at a positive camber angle
and with a trailing link suspension both inner and outer
wheels adopt camber angles equal to the roll angle. The loss
of cornering force can be seen from the typical curves shown
in Fig. 5.11.
As with the sliding pillar suspension, large camber changes
are anathema when low-profile tyres are used. The Porsche
Co. abandoned this type of suspension later in the 1960s.
11 12 13 14 4 5 9
9 2 3 8 7
Fig. 5.10 Parallel trailing link suspension on Porsche type 356B: 1,

front axle tubes; 2, torsion arm; 3, rubber buffer; 4, telescopic shock
absorbers; 5, tie rod; 6, brake drum; 7, adjusting nuts for front wheel
bearing; 8, opening for brake adjusting; 9, grease nipple; 10,
steering gear; 11, adjusting screw for sector shaft; 12, hexagon nut
for adjusting screw; 13, screw plug for oil-filling port; 14, coupling
disk; 15, torsion arm link pin; 16, stub axle (steering knuckle); 17,
steering arm at stub axle; 18, suspension arm link for stub axle.
2·0
_10°
1·8 -so
1-6 0°
1-4 + SO
z +10°
~
1·2
III
0
L
.2 1·0
01
C
·c
III
c
L 0·8
0
u
0·6
0·4
0·2
o 2 4 6 8 10
Slip angle (degrees)
Fig. 5.11 Influence of camber angle on cornering force.
5.5 Wishbone systems
5.5.1 Parallel links

Earlier wishbone systems usually had parallel links, often
of equal length as in Fig. 5.12. In this 1935 R-type MG
both front and rear suspensions were by parallel
double-wishbones. These wonderful little single-seater racing
cars skated round corners with all four wheels at astonishing
camber angles. The low mfM ratio helped to keep the tyre
Wishbone systems 91
Fig.5.12 The R-type MG Midget.
footprints glued to the road surface, but the loss of cornering

power from the positive camber angles (about 10% at 5° roll
angle) threw away. some of the inherent superiority of
independent suspension.
Tne live beam axles used by their rivals had small front
carltber angles and zero camber angles at the rear. At this
time, one must remember that the importance of camber
angle was not fully appreciated and the techniques by which
tyre makers now measure cornering force had not yet been
developed.
5.5.2 Angled double-wishbones

Under this heading we can consider true double-wishbone
systems or related systems such as the one popular on racing
cars in which a normal wishbone is used at the top and a
Fig. 5.13 Jim Clark's Formula Junior Lotus cornering at Oulton

Park. Note verticality of outer wheels.
single lower link. Sometimes the lower link is a single or

double longitudinal torque rod used to resist braking forces. If
we consider these three versions under one heading, they
represent one of the most popular layouts in use today.
There are four design parameters we can use to make
changes in the roll centre height and in the camber changes
and track variation with bump and rebound. These are:
(a) the relative angle between upper and lower links;
(b) the angle the lower link makes with the horizontal;
(c) the relative lengths of the upper and lower links;
(d) the ratio of lower link length to track.
The larger car manufacturers are able to use computer studies
to 'optimize' their suspension systems. As already stressed,
there is no perfect solution. The suspension engineer must
always choose a compromise. It is still possible, if one is
prepared to bum the midnight oil, for the young designer to
find a well-balanced design. Racing car designers have been
doing it for years.
Wishbone systems 93
Negligible change Negligible change in

in camber angle camber angle,
with side shiH
~.----
~--- Small track change under roll-----..,~

Fig. 5.14 Angled double-wishbone system for racing car.
For twenty years designers of racing cars and competition

sports cars have aimed at a suspension geometry that will
maintain verticality of the more heavily loaded outer wheel
under all angles of roll. Jim Clark's Formula Junior Lotus is
seen in Fig. 5.13 with both front and rear outer wheels very
close to the vertical in the middle of a tight bend. The camber
of the lightly loaded inner wheels is obviously regarded as
unimportant. Fig. 5.14 has been constructed to show that one
can juggle with the four parameters stated above and devise
an angled double-wishbone system which will maintain both
outer and inner wheels very close to the vertical, even at the
extreme roll angle of 10°. A typical racing car seldom exceeds
a roll angle of 1.5°.
5.5.3 No-roll layout

A pertinent question that must already have been asked by
many readers is: Why not choose a geometry that gives a roll
centre that coincides with the centroid? In this way roll would
be eliminated. Why not indeed!
We can only answer this question by asking another: Do
we really want to eliminate roll? It can be argued that roll to
the average driver is a rough indication of the extent of the
centrifugal force, but the medical profession has certain
reservations on the accuracy of this indication, particularly at
high values of g .
5.5.4 The physiology of roll

The human body was not designed (perhaps Charles Darwin
would dispute the use of this word) to interpret tum signals
from the antics of aeroplanes or automobiles. We try to
interpret the messages we receive through the middle ear, a
balance mechanism developed over millions of years to help
us to walk upright; certainly not to drive a car. The muscles
of the thighs and buttocks can be used to measure the
intensity of the side force produced by centrifugal force when
firmly held in a well-designed seat, but this again is not a
natural reaction. Early aviators complained that the parachute
upon which they were asked to sit turned them into 'deadend
kids', since this reduced their ability to sense the signals
transmitted to them by their buttocks.
Contrary to popular belief, the semi-circular canals in the
middle ear contribute very little to our sense of sideways
acceleration. The semi-circular canals are, of course, aware of
roll and many drivers react to the balance signals from these
canals by leaning away from the angle of roll. Underneath
these canals are two very important saucerlike disks in which
tiny hard particles are suspended in a jellylike substance.
These particles, called otoliths, collide with sensitive hairs
attached to nerve cells when the middle ear is subjected to
acceleration. These otolithic organs can' detect very small
deviations from the vertical, as demonstrated by so many
circus balancing acts. When a racing car is cornering, these
organs measure the roll angle plus the lateral acceleration. It
would, therefore, be difficult for the driver to differentiate
between these two factors and estimate his limiting cornering
speed if he had to rely on the otolithic organs alone.
Fortunately, if my analysis is correct, he also has the reaction
felt through the muscles of his thighs and buttocks. No
experimental evidence exists to support this hypothesis, but it
is interesting that all racing drivers insist on the necessity of a
well-tailored seat and their attendance at the works for a
fitting of the moulded seat contours is as much a ceremony as
Wishbone systems 95
Bertie Wooster's visit to his Savile Row tailor. With cornering
accelerations now reaching values of 2g, the strain on the
gluteous muscles of the upper thighs, on the back muscles
and certainly on the muscles of the neck are very high
indeed, so high perhaps that fatigue could begin to cloud a
driver's judgment towards the end of a race.
Over the years the roll angles on racing cars have become
less and less. Today they seldom approach 1.5° when
cornering at the limit. With such a small roll angle, the
guidance given to the driver from his otolithic sense organs
will be produced almost entirely by the centrifugal
acceleration. In the case of a racing car, therefore, the driver
has two senses to guide him when cornering. Modem family
saloons often roll at angles ranging from about 4° to as much
ZERO CAMBER
X
/'
, '
Fig. 5.15 Angled double-wishbone system with very high roll

centre. Effect of single-wheel bump.
as SO under lateral accelerations of O. 5g. The middle ear, with
its two sensing functions, must therefore be regarded as an
unreliable guide to the safe accelerations in a comer.
Experience of a particular model is still the safest way to build
up knowledge of the way a car handles in a comer.
No-roll suspensions have been the subject of several
interesting developments, as will be described in Chapter 10.
If we raise the roll centre on an angled double-wishbone
system until it coincides with the centroid of the sprung
mass, we now find that we have constructed a system that
gives undesirable changes in camber angle under single
wheel bump and rebound. This is illustrated in Fig. 5.15,
where both upper and lower wishbones are angled upwards
towards the car's centre line to give a high roll centre. In this
way one could construct a no-roll suspension, but the camber
changes in single wheel bump and rebound would be simply
unacceptable, as shown in the lower diagram.
5.5.5 MacPherson strut suspension

This system was invented by Earle MacPherson of General
Motors about thirty-five years ago. A good example of the
system is that fitted to the Datsun 260Z and 2S0Z sports cars.
A cross-sectional front elevation is given in Fig. 5.16. Upper
left is a plan view, giving a clear indication of the anti-roll bar
(stabilizer bar) and the compression rod, which resists
braking forces. Since the Datsun 260Z has rear wheel drive,
the compression rod is also under compression during
acceleration.
The construction of the MacPherson strut is typical. The
hub carrier is attached (by welding) to the outer tubular
member of the strut. At the base of this member is a sealed
ball joint mounted in the extremity of the lower transverse
link. Under maximum bump and rebound, the strut moves
through an angle of 2.5 0 and the required compliance is given
by a hollow rubber bush where the strut is attached to the
body. This rubber bush is bonded to a pressed steel housing
which contains a ball thrust bearing. Since the strut also acts
as a telescopic damper, the inner rod carries a damper piston
at its lower end. At the upper end of this rod a threaded
portion of reduced diameter is locked to the rubber mounting
Wishbone systems 97
I St ru t mounl in c Insulator 8 Wheel bea ring
2 Th rust beating 9 SUJpcn~ion ba H joint
) Bound bumper rubber \ 0 Tr:"nsvr=rsc lint
4 Dust CQWI I' Cum prcssiO'l1 rod
S Coil sprin& 12 StllbiUu. billr
6 Stru t a.u embly IJ Front $ u $pen~ion member
1 Front wheel hub
Fig. 5.16 MacPherson strut front suspension on Datsun 260Z.
and is located by the inner race of the ball bearing. This

thrust bearing allows free angular movement of the tubular
strut as the wheel is steered.
The Datsun is conventional in that the 'kingpin' offset is
positive. Negative offset or over centre-point steering, as shown
in Fig. 5.17, has been adopted by some designers of FWD
cars since it was first used on the 1966 Oldsmobile Toronado
to give improved stability under acceleration. Cars with dual.
circuit braking often have brake wheel cylinders
interconnected diagonally. Thus, if one braking circuit fails,
one front brake and the diagonally opposite rear brake is still
operative. With positive offset this type of failure creates a
torque tending to tum the vehicle around the front wheel
.....
~1IlllITl=_
I I
Fig. 5.17 Over centre-point steering.
with the effective brake. With hard braking this could result
in a spin. With negative offset the steering pull is in the
opposite direction, giving a better . chance of straight-line
braking under partial brake failure. A very useful bonus is a
reduction in the danger of spinning when a front tyre blows
out at speed.
S.6 Independent rear suspension
Most innovators meet a wall of resistance when they present

their brainchild; even worse in their eyes, they meet
indifference. Inevitably, when the new ideas have proved
Independent rear suspension 99
their worth, all the original objections evaporate, almost as if
they had never existed. We saw it happen with IFS. With
IRS, there has not yet been a complete victory. The
advantages are indisputable, but the live beam axle when
used at the rear is not as objectionable as when used at the
front. The high unsprung mass is the only real disadvantage.
Front wheel drive is becoming more popular. For the small
family car it is becoming the norm. General Motors have
designed a small 'world car' to be sold as the Astra by
Vauxhall Motors, as the Kadett by Opel and, no doubt,
under other badge guises in other countries. This follows the
current trend with a transverse front engine and FWD.
On FWD· cars therefore we see that the attractions of
independent suspension at the rear is faced with the sheer
simplicity of the dead beam axle, where the weight of a light
axle tube and simple locating links is only slightly heavier
than a good IRS system.
For a car of 1000 kg unladen weight, typical unsprung
weights for five popular systems would be:
1 Live rear axle with coil springs and kg/wheel
five-link location = 50
2 de Dion axle = 35
3 Double-wishbone IRS = 25
4 Semi-trailing arm IRS = 28
5 Dead rear axle = 32
Typical values for the unsprung/sprung mass (m/M) ratio for

the five systems would be:
System F/R weight distribution m/M
1 (RWD only) 50/50 0.2
2 (RWD only) 50/50 0.14
3(a) (FWD) 60/40 0.125
3(b) (RWD) 50/50 0.10
4(a) (FWD) 60/40 0.14
4(b) (RWD) 50/50 0.112
5 (FWD only) 60/40 0.16
For a RWD car the superiority of IRS is indisputable, for a ceu

with FWD the gain in road-holding appears to be marginal.
5.6.1 Wishbone IRS
The modem Jaguar rear suspension uses the drive shaft as an
upper suspension link, as shown in the cross section of Fig.
5.18. Twin coil springs and telescopic dampers are used, one
in front and the other behind the lower wishbone.
Fig. 5.18 Rear suspension on E-type Jaguar.

Independent rear suspension 101
Fig. 5.19 Lancia Flavia constant-velocity joint incorporating rolling

splines at inner end ofdrive shaft.
Longitudinal radius arms, not shown in the cross section, are

attached to the hub carrier and resist acceleration and braking
forces. By using a constant-length drive shaft, Jaguar cleverly
avoid one of the design problems associated with
double-wishbone IRS since sliding splines are not required.
Plain splines can exert coefficients of friction as high as 0.25
when under power. This interferes with the free working of
the suspension when under acceleration. Several early designs
suffered from this problem, but the bearing manufacturers
eventually provided an elegant solution. The Lancia Flavia of
1963 was an early example. Although this is a FWD car the
principle applies equally to the case of RWD. It will be seen
from Fig. 5.19 that the inner shaft universal joint uses rows
of balls that are free to slide in semi-circular grooves arranged
radially around the cylindrical housing. The outer shaft
joint is a constant-velocity joint of the normal Rzeppa type.
The Flavia used a transverse semi-elliptic spring which was
typical of many early IFS designs. Variations in interleaf
friction with this type of spring tended to sacrifice some of
the gains given by the roller bearing splines. Coil springs and
torsion bars have replaced semi-elliptic springs in modern
designs.
5.6.2 Semi-trailing arms
A swing axle gives negative camber on the outer wheel under
roll; a trailing arm gives positive camber. It is, therefore,
~o\\
~
B~ftRoll centre
SWING-AXLE
~ Roll centre
TRAILING LI NK
r
--
SEMI-TRAILING LINK
__ PLAN
~
I I ...............
::
~-=.::,.-=.-=.~ _ L - ____ ~
~ I 0
tW-
Effective swing-orm length I I
~oll
END ELEVATION
Effective
__ _ centresWing-arm
Fig. 5.20 Camber angle change under roll with swing axle, trailing
link and semi-trailing link suspensions.
Independent rear suspension. 103
~·~·,tm
~.II
'///
!O!315rh
I I
//'---\\\ ---t-.
I ~'"
n
~ I O'64m
I
O'65m ,~ . ! \ i, I O·965m
e=25_1_5_ _ rRear~alls _ \ I .
\ I
\~
~
Car
! Plan
1·42m
i
--~~----r-----
Axis and rear whLI centre line
, --------------- - '
Rear elevation
Fig. 5.21 Typical semi-trailing arm layout.
possible to combine them into a system called semi-trailing arm

or semi-trailing link to give negligible camber change under
roll. The concept was used on the Lancia Aurelia more than
twenty-five years ago. The development of the concept is
shown schematically in Fig. 5.20. The trail angle (t/J, in Fig.
5.21) is chosen by the designer to suit the dynamic
requirements of the particular vehicle. If for example he finds
it necessary to increase cornering forces at the rear to give
stable cornering, he could choose to increase the trail angle.
Several of the current General Motors cars use a trail angle of
26°. In the example shown in Fig. 5.21 the trail angle of
24°15' gives an effective swing arm centre that coincides with
the track. This layout would give a small amount of toe-in
(20' of toe-in for a bump or rebound of 100 mm). Since toe-in
/1~
Fig. 5.22 Ford Granada semi-trailing arm suspension.
is produced on both outer and inner wheels, the roll-steer

effect is negligible. When laying out a semi-trailing arm
system, the designer has many parameters to consider. There
is not only the obvious one of trail angle, there is the pivot
axis, which is parallel to the ground in Fig. 5.21, but can be
tilted if desired to change the static camber angle or can be
positioned above or below wheel centre height. Finally, for
every trail angle there is a· choice of effective swing arm
length: in Fig. 5.21 this is 1.42 m. If we increased this by
50% the roll centre height would be reduced by one-third and
the toe-in angle for 100-mm bump or rebound would be
reduced to about 13'. This would, of course, increase the
length of the semi-trailing arm by 50%. This could be difficult
to accommodat~ in the available space without intruding into
the passenger area.
The Ford Granada rear suspension is shown in Fig. 5.22.
Fabricated steel arms are used, pivoted on the cross-member
that supports the final drive unit. The halfshafts carry a new
design of constant-velocity sliding-type universal joint at both
ends.
References
[1] Olley, M. (1934), 'Independent wheel suspension - its whys

and wherefores', SAE Journal, March, p. 73.
[2] Goldman, E. D. and von Gierke, H. E. (1961), Effects of Shock
,and Vibration on Man, Shock and Vibration Handbook, vol. 3,
McGraw-Hill, New York, chapter 44, p. 441.
6
Road -holding
Many years ago we defined road-holding as 'the ability of a

car to follow the path dictated by an average driver in all
reasonable circumstances'. An average driver on snow or ice
will sometimes find that his car does not 'follow the path
dictated'. Snow and ice, to an average driver, are not
reasonable circumstances. To a Scandinavian rally driver, they
are. With his skill and experience he will usually persuade the
car to follow his wishes.
Depending upon the class of vehicle, we also demand the
ability to negotiate a bend safely within the limits of tyre
adhesion and the skill of the driver. Not only do we demand
higher cornering speeds from racing cars, but we assume
quicker reactions and greater skill from their drivers. The
reverse is assumed in the case of normal road vehicles. If the
driver should accidentally exceed the more modest cornering
power of the tyres, the car should be designed to be
'forgiving' within the possible limits of engineering
knowhow.
6.1 Straight-line stability
Oversteer and understeer were discussed briefly in Chap-

ter 4. For straight-line stability a degree of understeer is
essential. Let us consider a car travelling in a straight line
when it is suddenly subjected to a side thrust from a gust of
wind or from a dip in the road surface. Under the action of
this side force, all four wheels will run at slip angles tPl at the
front and tP r at the rear. If tP I is less than tP r' the car will steer
towards the disturbing force. The car will now be following a
curved path and since this will produce a centrifugal
acceleration the initial slip angles tP I and tP r will be increased
to new slip angles tP~ and tP~. The increment to tPl will
be less than that given to tPr' and this will again increase the
106 Road-holding
centrifugal force to a higher value. The car is obviously in an
unstable situation. This condition is called oversteer and, since
frequent steering corrections are required to hold a car that
oversteers in a straight line, it leads to rapid driver fatigue.
Stability can be produced by designing into the dynamic
behaviour of the car larger slip angles at the front than at the
rear under the action of transient side forces. With tPf greater
than tPr' a centrifugal force is produced which is opposed to
the side force. Within limits the car has intrinsic straight-line
stability. Such a car is not necessarily stable when cornering.
6.2 Cornering dynamics
When a car is cornering, a much higher centrifugal force is

produced than the minor side forces discussed in Section 6.1.
To balance this force the tyres generate side thrusts, and to
create these thrusts all four wheels must run at suitable slip
angles. As will be seen in Fig. 6.1(a), the longitudinal axis of
the car must be angled into the comer to achieve these
angles. Although the steering geometry may. be designed to
give an Ackermann centre based on the rear axle centre line,
the car could never tum about this centre since this would
give zero slip angles at the rear. For simplicity, we have
chosen to give identical steering angles to both inner and
outer front wheels. This not only simplifies the vector
diagrams in Fig. 6.1(b) and (c), but is a more logical steering
geometry than that proposed by Ackermann.
The true turning centre will depend upon the average slip
angles of all four tyres. In Fig. 6.1(a) an average slip angle of
7°30' has been chosen. In a typical understeering car this
would probably involve an average of 8° at the front and 7°
at the rear. As will be discussed later, one effect of body roll
when cornering is to transfer load from the inner wheels to
the outer. For any particular yaw angle, however, inner and
outer slip angles are fixed by the steering and suspension
geometry at the front and by the suspension geometry at the
rear. It is, therefore, not unreasonable to take an average for
the inner and outer slip angles in our simplified vector
diagrams.
Cornering dynamics 107
~yaWangle
~
__
(a)
Turning centre
t=~ --'-40
H4'
"
___ . ______-----------
~~ Ackermann
centre
1\
Sr
( b)
Tr
R·I
-Rr
- Rf
(e)
Rj
Fig. 6.1 Vector diagrams for cornering RWD and FWD medium-
size saloon.
We have chosen a rather tight turning circle in the example

of Fig. 6.1 to demonstrate the essential difference between
the cornering forces associated with RWD and FWD. The data
used in Fig. 6.1 represents a large European car or a
108 Road-holding
'compact' American car:
Total vehicle mass M = 1500 kg
Wheelbase = 2.8 m
Track = 1.6 m
Tum radius r = 10 m
Velocity v = 7 m/s.
From this, we obtain:
Mv 2
Centrifugal force C = -
r
1500 X 72
=----
10
= 7350 N.
This is a centrifugal force of about 0.5 gravity.
Fig. 6.1(a) is drawn to scale with a yaw angle of 7°30'. This
yaw angle is necessary to promote the required slip angles.
The turning radius is perpendicular to this instantaneous yaw
axis and the turning centre is at O.
6.3 Rear wheel drive
In the Circle of Forces or closed vector diagram for the RWD

car in Fig. 6.1(b) the centrifugal force C of 7350 N is balanced
by the following forces:
Sri the cornering force from the rear tyres;
Trf the tractive force from the rear tyres;
Sfl the cornering force from the front tyres;
Ril the inertia force required to produce acceleration of the
mass through the bend;
Rfl the road resistance of the front wheels;
Rrl the road resistance of the rear wheels.
If we refer to our earlier concept of a tyre-footprint Circle of
Forces (see Fig. 1.17 in Chapter 1), we see that Tr and Sr can
be found by the following relationship:
Total tyre thrust = J(S; + T;).

Front wheel drive 109
If we assume for simplicity that the front and rear tyres exert
identical total thrusts
R j, the inertia force, is that force required to accelerate the

mass through the bend:
Rj=Ma
where a is the acceleration. In the particular example a value
for a of 1.0 m/s 2 has been taken.
Therefore Rj = 1500 N.
Although the speed is low, the bend radius is enough to
increase the road resistance to an appreciable value. A value
of 75 N per wheel seems reasonable. No great accuracy is
required in estimating this value. If the values Rr and R f of
150 N for each pair of wheels were neglected, the error
would. not be great.
Only C, R j , R f and Rr are known in direction and
magnitude. TTl Sr and Sf are known in direction only, but the
'relationship Sf = V(T; + S~) helps us to construct a closed
vector diagram by trial and error, as shown in Fig. 6.1(b).
6.4 Front wheel drive'
The same technique has been used to construct a closed

vector diagram for the case of a FWD qu. It is immediately
apparent that the direction of the tractive effort in this case
has a centripetal component that opposes a percentage of the
centrifugal force.
Scaling off values from Fig. 6.1(b) and (c), we get the
following comparison:
RWD FWD
4700 N 2200 N
2650 N 4050 N
3950 N
3400 N
A direct comparison of the cornering 'efficiency' of the two
110 Road-holding
systems can be obtained by comparing the value of Sf in Fig.
6.l(b) and Sr in Fig. 6.l(c). This shows a reduction in
cornering forces of about 15% in the second case when
resisting the same centrifugal forces. For the same limiting
slip angles the FWD car should be able to negotiate such a
tight corner at about 7% higher speed. This. gain falls to about
3% for a 20-m radius corner and becomes negligible for large
radius curves.
6.5 The cornering racing car
Modern racing cars carry such a high percentage of the

weight and the footprint area at the rear that they can be said
to be steered in a bend by the throttle foot. A typical weight
distribution would be 33/67 and typical tyre widths 20 in at
the rear and 10 in at the front. Rear end loads are usually
increased in all but the most acute bends by the downthrust
of a rear wing, a large inverted aero foil. A smaller front wing
is provided to increase the effective load on the front wheels.
During 1978-9 the overall downthrust was increased beyond
the actual weight of the car. This was achieved by the
application of a Venturi-effect underneath the car and the
development of sliding skirts on the body sides to Jseal-in' the
depression below the car. The penalty of these aerodynamic
tricks has been a loss in maximum speed, but the cornering
speed has been increased dramatically. Measurements
reported by Motor (3 March 1979) made on a Lotus 79 gave
cornering accelerations of 2.05g. Fig. 6.2 shows how the
downforces are distributed on the Lotus 79.
Fig. 6.2 Aerodynamic downforces on Fl racing car at 150 m. p.h.

The cornering racing car 111
12·5° Yaw angle Turning
l . _________
centre0
1 ---- ----- ---

\ I' ----- - ----- ------
~~-~- -~~-~-~~:=-=---~
~-- Ackermann
i\ (a) centre
s,
5;"-----
---
(b)
Fig. 6.3 (a) and (b): Vector diagram for cornering racing car.
6.5.1 Cornering dynamics

Fig 6.3(a) shows a typical Formula 1 car at a yaw angle of
12°. The following data have been used to construct the
vector diagram of Fig. 6.3(b):
Total vehicle mass M = 725 kg
Wheelbase = 2.7m
Track = 1.6 m
Weight distribution FIR = 33/67
Tum radiusr = 50m
Velocity v = 30 mls
Therefore,
Centrifugal force C
725 X 30 2
50
= 13050 N.
This is a centrifugal force of about 1.8g.
112 Road-holding
In Fig. 6.3(b) the broken lines e', T' and S~ show how the
driver can lift his throttle foot, reduce the torque from the rear
wheels and change the turning centre. This, of course, is a
steering action. The changes in Sf and Sr are negligible for a
large change in direction, giving the car good stability in a
comer.
6.6 Yaw inertia'
In Chapter 3 the polar moment of inertia about a horizontal axis

through the centroid was considered. This has an influence
on the natural period in pitch. The disposition of the major
/ --- ----
'1- ---, - If- J.--'t1--;r---
, ____ !..." ~ )0",
______ • ~
,- - - ,~
,.,.
__ -
-~.
\ .. I. \
--
1
'--I) I -\-,_; ...... _ - _ .... ', 1 I I
1 I: --- )..l. I~ __ J
: ::,' r---"'_ ,.- ... ~I-V' ,----.I
I
-- --.l- --' -,: -:. :- ,. .).: --
I J "... ,
l
,L _ _ _ .!j'" __ .. ..:.-:, ... --~.!'-.-'_~-;---
....
I I,~'
'T,' I
, _ _ _ _I 1_ I
,
'''-.,
.... --- ----- /
-
(a) HIGH POLAR MOMENT OF INERTIA
1-----1 , - ' 1-----1

, 1 ~-r-J_I I
1_ _ _ _ _ 1 /" " 1__ 1........ I
-"",---
-----7~·-
~_- - - ' , ) ,I
/
, ---, \
~
\ _ \ - ..... _ _ ,
:- -=-=-== .... \
,
I
. -
---=L
-- __ "
' I
.---+
"-. --- t - - - - I
:-- - - -! -- -M 1 1
1-----1 ' __I L _____'
(b) LOW POLAR MOMENT OF INERTIA

Fig. 6.4 How the distribution of the major masses influences the
radius of gyration.
Yaw inertia 113
masses in the car also influences the resistance of the vehicle
to change in direction. This polar moment of inertia is that of
the whole vehicle, including the unsprung components and is
a measure of the car's resistance to rotation about a vertical
axis through the centroid. Since this rotation involves a
change in yaw angle, it is sometimes called the yaw moment of
inertia.
Racing drivers discovered many years ago that
long-wheelbase versions of the same basic racing car were
much more stable at speed than the short-wheelbase model.
This was before the days of aerodynamics. When negotiating
a twisting part of a round-the-houses circuit, they preferred
the handling of the short-wheelbase version. Fig. 6.4 shows
how a front-engine rear drive car, carrying a full complement
of passengers, will have a much higher yaw moment of
inertia than a mid-engine sports car with only the driver
aboard.
The Moment of Inertia in Yaw is given by:
I Y =Mv 2
t''-y (6.1)
where M t is the total mass, and Ky is the radius gyration of the
complete car. The couple, to produce any change in direction,
must come from the tyre footprints acting through effective
arms of length 11 at the front and 12 at the rear.
If we consider the vehicle as changing from the
straight-ahead direction to an angular velocity of w rad/s in t
seconds, the angular acceleration a = wit rad/s 2.
The tyre forces to produce this angular acceleration are:
""~ s ---.!L
- 11 + 12 . (6.2)
In the example of Fig. 6.1 the angular velocity
v 7
w = r= 10 = 0.7 rad/s.
If this rate of angular velocity is achieved in 1 s, the angular

acceleration would be 0.7 rad/s 2.
A typical value of Iy for a medium-size saloon of 1500 kg
would be 2500 kg/m2. The summation of the tyre forces
114 Road-holding
(inwards at the front and outwards at the rear) would be:
L s = 2S00 x 0. 7 = 620 N.
2.8
Dividing this equally between front and rear in the example

chosen for the vector diagram in Fig. 6.1(b), the front tyre
forces would be increased from 4700 N to SOlO N and the
rear tyre forces reduced from 26S0 N to 2340 N. These are
only transient forces induced during the 1 s duration while
the car is steered from the straight-ahead direction into the
10-m radius curve of Fig. 6.1(a). Since this only represents a
centrifugal force of about O.Sg, the transient angular
acceleration would be easily accommodated on a dry road by
an increase in the front slip angles and a reduction at the rear.
On a wet road the sudden application of O.Sg could provoke
a front wheel skid.
The above analysis of the influence of yaw inertia neglects
the effects of roll-steer and load transfer from inner to outer
tyres under the effects of roll. Some modern sports cars
introduce another variable into this complex situation. The
Lotus Esprit, for example, uses 20S/60 HR14 tyres at the front
at a pressure of 18 p.s.i. (124 kN/m 2) and 20S/70 HR14 at the
rear at a pressure of 28 p.s.i. (193 kN/m 2). The 60 profile
tyres at the front have a larger footprint area than the 70
profile tyres at the rear and will, therefore, transmit greater
side forces for a given slip angle.
6.6.1 Spin or slide?

If the driver of a very large four-wheeled vehicle such as a
bus tries to change direction too quickly on a slippery surface,
the breakaway usually occurs at the front end first. Once the
vehicle is sliding, however, the angular momentum that
provoked the slide usually results in a spin; the larger the
ratio K~/(ll x 12 ), the more prolonged the spin. A small car
like the Volkswagen Beetle with a tail-heavy weight
distribution usually begins to slide at the rear under similar
circumstances and a spin is inevitable with a driver of less
than average ability or even a good driver who has been
behind the wheel for too long.
Fig. 6.5 Relative yaw responses (on left) and relative yaw inertias
(on right) for eight possible mass distributions on a modern sports
car.
116 Road-holding
6.6.2 Yaw responses

Measurements have been made by various research
institutions of the influence of yaw inertia on the response of
a car to a given steering input. Yaw response is usually taken
as the lag between the input from the steering wheel and the
output in terms of change in angular momentum.
Fig. 6.5 is based on a series of calculations made by the
Jaguar Car Co. and presented in a paper to the SAE Detroit
Congress (Feb.-Mar., 1977) by Bob Knight in conjunction
with Jim Randle. The particular study was made as a
justification for the layout chosen for the XJ-S. Not that it
needs any! In Bob Knight's comparison yaw response is the
inverse of the above definition. It is the rate of change of
angular momentum for a given change in steering angle. In
Fig. 6.5, (a) is taken as unity. A car with a higher value is,
therefore, one which in common language 'responds more
quickly to the helm'. None of the examples can be said to
represent the XJ-S which has a weight distribution of 57/43
and has a different profile to those of (a) and (b).
As one would expect, the poorest yaw response is given
by (g). Not only is the value of I y 8% higher than that of (a),
but the weight distribution of 61/39 will obviously make the
front slip angles increase to a greater extent for a given
steering angle change than would occur with a tail-heavy car.
In contrast we see that the tail-heavy (e) with a relatively low
yaw inertia of 0.94 has the highest relative yaw response.
With a weight distribution of 41/59, however, great care will
be necessary in the layout of the suspension system to avoid
the dangers of terminal oversteer when cornering near the
limit.
The above analysis suggests that the range of yaw
responses for a comprehensive list of modem sports car
designs is not large. If we extended the survey to cover
everything from a small sports car such as an MG Midget to a
full-size American sedan, we would find the range of yaw
responses to be much greater; certainly greater than 2 to 1.
6.6.3 Spin accelerations

A recent attempt by the staff of Motor, with the help of the
Cranfield School of Automotive Studies, to measure the
Roll-steer 117
maximum spin accelerations that could be reached .on a
wetted skid pad produced some conflicting results. A Lotus
Elite, with a relatively high yaw moment of inertia of
2006 kglm 2 and a value for K~/(ll x [2) of 0.96, gave an almost
identical spin acceleration to the mid-engined Esprit for which
the corresponding values are 1508 kglm 2 and 0.83. The Elite
has 205/60 tyres all round, while the Esprit has 205/70 at the
front and the wider 205/60 at the rear. Such differences as
profile ratios are obviously more important than differences in
yaw inertias.
6.7 Roll-steer
When a wheel is deflected from its normal axis of rotation by
body roll, it can influence the car's direction. When both
wheels at the front or both wheels at the rear are deflected by
roll, the roll-steer effect can be quite pronounced. It all
depends upon the suspension geometry.
An example of roll-steer can be illustrated by the simple
example of a dead rear axle (see Fig. 6.6) as used on many
FWD cars in which the trailing links slope upwards to the
rear. As the body rolls, the more heavily loaded outer wheel
moves upwards and fonvards while the inner wheel moves
downwards and backwards. This produces understeer. In
practice the link pivot centre would be only slightly below
wheel centre height, probably no more than 50 mm. The
same technique can be used with a semi-trailing link
suspension or with any independent suspension design using
longitudinal links to control the fore and aft location of the
wheels. It can be applied to both front or rear suspensions.
Excessive roll oversteer can make a car 'twitchy' on an
uneven road surface. A memorable example was the original
Mercedes-Benz 300SL with double-wishbone suspension at
the front and swing axle suspension at the rear. The low roll
centre at the front, the very high roll centre at the rear, plus
the variable cornering power produced by single wheel bump
and rebound at the rear as the rear wheels alternate between
positive and negative camber, all contribute to a lack of
stability of speed. (I myself remember this tendency to weave
from side to side at high speed!) Daimler-Benz removed this
inherent instability when they introduced the low-pivot
swing axle design.
118 Road-holding
, I
Fig. 6.6 Deflection steer: roll understeer on a dead rear axle.
6.7.1 Deflection steer

On production cars it is common practice to use rubber
bushes at the suspension pivots. With metal-to-metal contact
far too much harsh road vibration would be transmitted to
the occupants. Unfortunately, the deflections that can occur
under changing loads applied to these flexible pivots can
produce undesirable changes in the suspension and steering
geometry. Even variations introduced by production
tolerances, can give problems. The original Ford Escort was
designed to have zero roll-steer on the front wheels. In
practice some cars leaving the a_ssembly line toed-in under
Roll-steer 119
Toe-ou~ \
for \
------- --
undersffier~
---
----- ---- --
~----
_ - .. 0
Toe-our
for ~
overs~eer
. --- --- - - - - --- --- --- --- -'---.
-- -
Fig. 6.7 Effects of toe-out deflection changes at front and rear.
bump (oversteer), while others toed:-out (understeer). When

the front suspension was modified in 1968, the geometry was
changed to give roll understeer at the front on all cars. The
designed roll understeer was 13 minutes per degree of roll.
This would be greater on some cars and less on others, but
the manufacturing tolerances were such that no cars would
reach the customer with a tendency to roll oversteer.
Deflection steer can occur at either end of the car. As
shown in Fig. 6.7, toe-out of the outer wheel under roll
produces understeer at the front but oversteer at the rear. The
Porsche engineers who designed the type 928 tackled the
problem of deflection steer in a novel way. This patented
'Weissach-Axle' is described in Chapter 12, in which we
present a detailed analysis of the type 928 suspension.
7
Dampers
7.1 Friction dampers
Originally called shock absorbers, dampers were first introduced

in the 1920s as a palliative to reduce excessive movements of
the front axle. The popular Andre-Hartford friction damper,
shown in Fig. 7.1, uses a sandwich arrangement of alternating
flat disks of steel and friction material, the whole assembly
tightened by an adjustable dished spring. Two levers in
V-formation, pivotally attached to the axle and the chassis
frame, produce relative angular movement between the steel
disks and the friction disks, thus providing a damping action.
A serious disadvantage of the friction damper was a
tendency to exert high initial or starting friction, i.e. more
friction was required to create initial movement than was
exerted when moving. The model shown in Fig. 7.1 is the
Telecontrol adjustable shock absorber. In this design the
pressure on the friction disks could be varied by means of a
Bowden cable controlled by a lever on the dashboard.
7.2 Hydraulic dampers
A simple hydraulic damper can be made from a piston

working inside a cylinder with oil on both sides of the piston.
Fig. 7.1 Andre-Hartford Telecontrol shock absorber.

Hydraulic dampers 121
Fig. 7.2 The principle of the 'ride control' or 'linear valve'.
A series of holes in this piston and a spring-loaded valve allow

oil to flow from one side of the piston to the other, the size
and number of holes and the design details of the valve
determine the forces exerted by the damper on the
suspension system. The piston is usually attached to the
Bleed Linear valve opening
I ._
. .J-"'---
Burnp"a\~2.---- I
J...---
,',
I
-.-
.--
I
,
~' , I
",,' I I
Speed
Fig. 7.3 Typical flow characteristic curves given by bleed flow and
linear valve flow.
122 Dampers
body, the cylinder to the unsprung mass. It is easy to see that
the use of a viscous fluid in the damper will give a damping
force that is 'velocity-dependent', i.e. F ex: V n • The value of the
index n will depend upon the design of the spring-loaded
valve. At very low vertical wheel velocities the valve is
designed to remain closed, all oil flow being through the
bleed orifices. These are small enough to give streamline
flow. The modern 'ride-control' or 'linear' valve is
spring-loaded, as shown in Fig. 7.2. By changes in spring
strength and the diameter and port dimensions of the valve
orifice, a wide range of resistance/flow curves are possible. As
the linear valve opens under high fluid flows, a point is
eventually reached when the valve disk is lifted so high that
the flow is controlled by the diameter of the valve orifice.
This gives a gradually increasing rate of rise in resistance, as
seen in the rebound valving curve of Fig. 7.3.
Most damper designers provide different valves for bump
and rebound flow, the resistance usually being much less
in bump than in rebound. Typical characteristic resistance
curves are given in Fig. 7.3.
7.3 The double-tube damper
Older designs of telescopic damper used two concentric

tubes. The inner tube was the working cylinder, the outer
tube acting as a reservoir to hold the excess oil moved on the
downstroke. A good example is the Koni, shown in cross
section in Fig. 7.4. Since liquids are almost incompressible,
the necessity for the outer chamber is apparent when it is
seen that the effective area on the upstroke is less than that
on the downstroke by the cross-sectional area of the piston
rod.
7.4 The single-tube damper
The Girling Monitube damper uses a principle first used by

de Carbon nearly fifty years ago. With modern shaft-sealing
The single-tube damper 123
Seals
Rod
Cy li nder
Reservo ir
Non-return
V al ve
Piston
Cal ibrated
Cha nne l s
By·pass
Valve
Foot
Valve
Fig. 7.4 Koni double-tube damper.
techniques, Girling has produced a successful development of

this old idea. The recuperation cylinder in this damper is an
extension of the working cylinder containing gas under high
pressure. A lightweight glass-reinforced, nylon-free piston is
used to isolate the two chambers. Movement of this light
piston accommodates the displacement differences of the
working piston. Valves for both bump and rebound are
incorporated in the working cylinder, but are omitted from
the simplified diagram in the drawing on the left of Fig. 7.5.
Another ingenious single-tube damper has been developed
by Woodhead-Munroe, shown in principle on the right of
124 Dampers
• ". -, .1
'.
. ~ . .
00 0
o
o
o
o 0
o 0
o
Fig. 7.5 Girling Monitube principle (left) and Woodhead-Munroe

Monotube principle (right).
Fig. 7.5. Aeration in earlier designs of hydraulic damper

sometimes occurred when they had been subjected to
heavy-duty work, such as a long drive over rough roads. The
effect of aeration was similar to a large reduction in oil
viscosity with a corresponding drop in the effectiveness of the
dampers. The Woodhead-Munroe engineers decided that the
most elegant solution to the problem was to keep the oil
aerated at all times. By charging the damper cylinder with a
finely emulsified mixture of oil and gas, they also solved the
problem of the differing swept volumes on bump and
re bound, since the variations in displacement could be
absorbed by compression of the tiny gas bubbles. The
valving had to be designed to suit the liquid/gas mixture,
since this had a relatively low viscosity, but this had now
become an effective viscosity that varied very little from the
start to the end of a journey. A cross section ofa
Woodhead-Munroe monotube damper is given in Fig. 7.6.
Damping theory 125
BUMP R<9OUND
0.
Fig. 7.6 Woodhead-Munroe Monotube damper in cross section.
7.S Damping theory
Although our aim is to damp out as quickly as possible each

vertical impulse given to the wheels, a measure of com-
promise is necessary. Excessive damping gives a reduc-
tion in the initial spring deflection. This gives a harder ride.
126 Dampers
C..c.,.f(4kIM)
Cc =.f(4kIM)
C >!c4kIM)
,
11/
>
Fig. 7.7 Damping with various damping coefficients.
7.5.1 Critical damping

The concept of a critical damping coefficient is a purely
mathematical one and must not be confused with the more
practical concept of optimum damping.
Critical damping is given when the damping force is the
minimum force required to prevent the initial displacement of
the sprung mass from the equilibrium position, resulting in
an oscillation that crosses the equilibrium position. Critical
damping is indicated in Fig. 7.7 by the curve labelled C e •
The critical damping coefficient is given by:
C=J4kM
e
(7.1)
where k = the spring rate (N/m) and M = the sprung mass

(kg). The damping coefficient has the units of frequency, i.e.
l/s.
The mathematics involved in establishing the above
equation are very involved. Students are referred to Mechanics
of Road Vehicles by W. Steeds [1] for confirmation. Critical
damping, in· any case, is much too severe for a practical
suspension system, but it does give us a yardstick by which
we can express the degree of damping given to a practical
system. On this scale most modem automobiles are given
damping coefficients C where ClC e = 0.25 to 0.5.
A typical modem damper would damp out the initial bump
Damping theory 127
Fig. 7.8 Attenuation curve for typical damper.
deflection of x to a value of no more than O.2x by the second

bump in a time interval of IIF n seconds, where F n is the
natural frequency of the suspension. This is illustrated in
Fig. 7.8.
7.5.2 Resonance
We have already referred to the phenomenon of resonance in
Chapter 2. As shown in Fig. 2.6, with a complete lack of
damping the effect of a regular wave formation in the road
surface could be quite shattering when the forcing frequency
ir coincides with the natural spring frequency in. Curve 1
demonstrates this. Curve 2 with a maximum acceleration at
resonance of about 0.7g, has a value of CjC c = 0.15. Curve 3
is for a value of CIC c = 0.25 and Curve 4 is for CIC c = 0.65.
Curve 3 has been labelled 'well-damped', but these curves
are based on a damping function directly proportional to
vertical wheel velocity. Damper designers have accumulated a
wealth of practical experience in the design of ride-control
valves that gives a finished product giving good damping
over a very wide range of frli n values.
7.5.3 A mathematical model

Before we can hope to use computers to design suspension
systems, we must first establish reliable mathematical models.
128 Dampers
This was an irresistible challenge to academic minds, even
before computers were available. As early as 1950 de Carbon
presented a paper to the Congres Technique Intemationale de
l'Automo bile in Paris on the theory and practice of vehicular
suspension damping. After thirty years of mathematical
analysis, reasonably accurate models can be constructed to
help the design engineer through his computer facilities to
optimize the suspension damping on a new model. (One
textbook and four useful technical papers, [2], [3], [4], and
[5], to assist the serious student are given in the References.)
References
[1] Steeds, W. (1960) Mechanics of Road Vehicles, Iliffe & Sons,

London.
[2] Bender, Kamopp and Paul (1967), 'On the optimization of
vehicle suspensions using random process theory', Proc. 1967
Transportation Engng Conf., ASME paper 67-trans-12.
[3] Thompson, A. G. (1969-70), 'Optimum damping in a randomly
excited non-linear suspension', Proc. Inst. Mech. Engrs, (AD)
1969-70, vol. 184, pt. 2A, no. 8, pp. 169-84.
[4] Thompson, A. G. (1972), 'A simple formula for optimum
suspension damping', J. Automotive Engng, Inst. Mech. Engrs,
April.
[5] Hever, P. J. (1969), 'Interaction of suspension and vibration
damping in road vehicles', ATZ, 71, (12) (December), pp. 446-51
(in German).
8
Pneumatic Suspensions
8.1 The air spring
A mechanism for compressing and expanding an enclosed

volume of air by means of a piston or diaphragm was first
patented as a suspension spring in 1906 and a Cowey car
fitted with pneumatic suspension was exhibited at Olympia
Motor Show in 1909.
To demonstrate the concept we can compare the behaviour
of a conventional steel spring, such as a coil spring or a
torsion bar and that of a pneumatic strut, shown
schematically in Fig. 8.1. Friction in the strut will be
M= 400 kg
TIV
250mm
11----;--11
Fig. 8.1 Elementary pneumatic strut.

130 Pneumatic suspensions
10·0
9·0
8'0
~ 7·0
z
~
Q/ 6'0
~
.2
] 5'0
I..
~
4·0
3·0
2·0
1·0
o 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150

Deflection in bounce ( mm )
Fig. 8.2 Comparative compression forces with normal coil spring
and pneumatic spring.
neglected. Let us consider the mass supported by the spring

as 400 kg. With a rate of 20 kN/m a coil spring of constant
gauge, coil diameter and helix angle will exert a restoring
force of 0.5 kN for a deflection of 25 mm, 1.0 kN for a
deflection of 50 mm and pro rata up to a value of 3.0 kN for a
deflection of 150 mm, as shown in Fig. 8.2.
Very slow compression of the air in the strut would give a
rate curve following the law of isothermal compression, i.e.
PV = a constant. At speed however on a typical road surface,
the frequencies imposed on the suspension system, omitting
those absorbed by the tyre, vary from zero to about 40 Hz.
Self-levelling systems 131
Compression and expansion in general will, therefore, be
very rapid and will be much closer to adiabatic. Since some
cooling will inevitably occur, polytropic compression and
expansion is assumed with an index of n = 1.33. This curve
in Fig. 8.2 is, therefore representative of the compression
curve for a simple pneumatic strut where PV1.33 = C.
From Fig. 8.2 we see that the spring rates for small
deflections are fairly close, but for large deflections the
pneumatic spring has a much greater rate. The extent of this
increase in rate can be varied by increasing or decreasing
the strut length in relation to the maximum deflection.
Alternatively, an auxiliary chamber can be provided to
increase the compressed volume at maximum bump.
8.2 Temperature variations
A simple pneumatic strut would give a variable static ride

height. If the ride height was at the designed height of
Fig. 8.1 at 20°C, it would fall by 34 mm if the car stood
overnight at a temperature of - 20°C. More devastating
would be the effect of a rise in temperature from compression
heating as the vehicle travelled on a rough road. A calculation
made in Automobile Engineer in March 1955 suggested a rise of
about 65°C in the air temperature without the provision of
cooling fins or a forced cooling system. With an ambient
temperature of 35°C this would increase the ride height by
68 mm, an unacceptable increase.
8.3 Self-levelling systems
Variations in ride height is something we have lived with for

so long that we accept it as normal. Consider a typical family
saloon with a load on each rear spring of 300 kg with an
empty rear seat and nothing in the boot (trunk). With a
typical rear spring rate of 20 kN/m and an available wheel
travel in bump of 120 mm (4.7 in) the rear ground clearance
could be as shown in Fig. 8.3, i.e. 175 mm (6.9 in). An
increase in load of three passengers at the rear and 50 kg of
luggage would increase the load on each rear spring by a bout
~- - -
Max . bump ~ 120 mm

J:. __ _
Fig. 8.3
140 kg. The additional downforce on each rear spring would

be 140 x 9.807 = 1373 N, giving an additional static spring
compression of 69 mm (2.7 in). The static ground clearance at
the rear would be reduced to 106 mm (4.2 in) and the
available spring travel in bump reduced from 120 mm (4.7 in)
to 51 mm (2.0 in). On a rough road the rear springs would
spend much of the time hitting the bump stops.
A self-levelling system is an obvious answer to carrying a
wide range of loads. What is of no less importance is the
freedom it gives to the designer to soften the suspension.
With a bump travel of 120 mm available under all load
conditions the spring rate can be softened from 20 kN/m to as
low as 12 kN/m. This would reduce the rear spring
frequencies from the typical conventional values of 1.3 Hz
(2-up) and 1.07 Hz (5-up) to 1.0 Hz (2-up) and 0.83 Hz
(5-up).
8.3.1 Practical air suspensions

The Firestone Tire Co. of America experimented in the 1930s
with bellows springs, using two convolutions of the type
.1
BEA.O RING I
I STEEL BEA.O ,/
l PLIES NYLON CORO
OIL RESISTING RUBBER
Fig. 8.4 Section through two-convolution air spring.
shown in Fig. 8.4. It was a rubber and fabric moulding with

a steel bead ring at each end and a steel girdle ring to resist
expansion in the middle. When the upper and lower bead
rings were bolted to an upper steel plate attached to the
body, and a lower steel plate mounted on the lower
suspension arm, an enclosed chamber was formed to act as
an air spring.
From these early experiments the company developed an
air suspension system which was fitted to long-distance
Greyhound coaches in the 1950s. In this installation the
bellows unit was designed to operate at a pressure of 4.5-5.2
bar (65-75Ibf/in 2). Since commercial-vehicle air brakes in the
USA are designed to operate at slightly higher pressures, this
made the system suitable for use with the standard air brake
compressor. Natural frequencies in the range 1.3-1.5 Hz
could be attained using this system. These were very low
frequencies for a large bus. Since the bellows unit was
frictionless, apart from the negligible amount of hysteresis in
the moulded bellows, it was necessary to provide good
double-acting dampers and bump stops to cope with the
worst roads.
In the Greyhound installation each suspension unit
consisted of a platform mounted above the axle with a
bellows unit fore and aft. A surge tank was installed above
each pair of bellows. The bellows had no inherent stability,
but were located longitudinally by radius rods and laterally
by a Panhard rod. The total volume of the air spring was that
of the two bellows units plus the capacity of the surge tank,
since an unrestricted flow was permitted between the three
units. The action of the surge tank was to control the spring
rate and the natural frequency, as will be described later.
From each surge tank a pipe passed to a levelling valve,
there being one levelling valve per wheel. The levelling
valves were pressurized from the main storage tank which
was in series with the brake system storage tank. A check
valve, fitted between the two tanks, ensured an air supply to
the brakes if the suspension system failed. The levelling
valves maintained a constant ride height on all wheels
irrespective of the static load. The valve body was secured to
TO AIR SPRING
AIR INLET
FROM
STORAGE
TANK
Fig. 8.5 Cross section of levelling valve.

the chassis frame, the valve actuating arm to the axle. An
increase in load moved the arm inwards (see Fig. 8.5) which in
tum tilted the actuating rod to open the inlet valve and admit
compressed air to the air spring. Conversely, with a reduction
in load the actuating rod would tilt in the opposite direction
to exhaust air from the spring until the actuating arm became
level again. Valve operation was not instantaneous, however,
since a time-delay mechanism was built into the system. The
end of the actuating arm acted on the rod through balanced
compression springs contained in a central passage in the rod.
These in tum were attached by vertical links to two hydraulic
dampers. The speed of movement of the actuating rod was
controlled entirely by the time-delay bleed between the two
dampers. With a sudden movement of the actuating arm in
either direction tilting of the rod only occurred after several
I
10
I
9
I
8
x 6
c ,/
/
~
I 5
~
'" f
/ B"" ""
1
~. V .... "'"~"'" .... ...-
-.......:: ~. 4..: ~-
o 1 1 1 4 5 6 7
STROKE-in.
EXTENDED COMPRESSED
Curve A No surge tank.

Curve B - - - - 1770 in3 surge tank.
Curve C - - - - - 3540 in 3 surge tank.
Fig. 8.6 Effect of surge tank volume on spring rate.
110
'"~ ~
E
~'OO
>-
u
z
w
::I
:Z90
---r----
:f
"fso
Z r--..
'"
70 0 10 20 30 4OxI0 2
SURGE TANK VOLUME (inl)
Fig. 8.7 Effect of surge tank on volume spring frequency.
seconds, the actual time being controlled by the time taken

for the appropriate compression spring to overcome the
resistance to flow through the orifice.
The time delay was essential to desensitize the levelling
valves. With no time delay the valves would open and close
under the action of every bump and rebound movement. Not
only would this put a high demand on the air supply and
require the provision of a very large compressor, but the
concept of four individual air springs would be upset and
fore-and-aft and side-to-side interaction could occur.
The actual delay time was established on the Greyhound
buses by trial and error. It was later found entirely
satisfactory to fit two levelling valves to the rear axle and a
single valve at the front to control both front springs. The
valve was designed to prevent any transfer of air from side to
side.
Without a surge tank the spring rate increase with
increasing wheel movement would have been far too drastic,
as is shown in Fig. 8.2. On such a large vehicle a'large surge
tank could easily be accommodated, but a large-bore
connecting pipe was needed to ensure that the two bellows
and their surge tank would behave as if they were a single
large volume. Fig. 8.6 shows how spring rate changes could
be reduced by the addition of a surge tank. Fig. 8.7 shows
how spring frequency is reduced as the surge tank volume is
increased.
Pedestal type 1 Pedestal type 2

Fig. 8.8 Comparison of pedestal shapes.
t
"'EBOUND
SPRING STROKE
BUMP
Fig. 8.9 Dynamic load curves using pedestals of different proffies.
8.3.2 Single-convolution air springs
With the automobile market in view the Firestone Co.
developed a rubber and canvas bellows with only one
convolution. To vary the spring rate characteristics, curved or
tapered pedestals were developed to give an effective piston
area that changed throughout the stroke. Fig. 8.8 shows two
such pedestals, and Fig. 8.9 shows how a change in pedestal
profile can influence the load/deflection curve. With pedestal
2 soft suspension will be given for small deflection in both
bump and rebound (curve BC). For greater deflections in
bump the spring operates over the rising section (CD). In
rebound the spring operates over a more modest rising rate
curve (BA).
8.4 The General Motors air suspension
The Eldorado Brougham was introduced in 1957 with air

suspension by the Cadillac Division of General Motors.
Despite the excellent ride and the advantages of self-levelling,
the system was replaced by conventional steel coil springs in
1960. Problems with air leaks in some components and
frozen valves in subzero conditions would undoubtedly have
been solved by the Cadillac engineers, given time, but the
Fig. S.10 Cross section of Cadillac front air spring.

The General Motors air suspension 139
Fig. 8.11 Cross section of Cadillac rear air spring.
proven reliability of coil springs was an inexpensive and quick

solution to an embarrassing situation. It was admitted by
Motor that the earlier Cadillacs had set a very high standard.
Even so they stated that the air-sprung Eldorado 'gave an
amazing improvement' over badly surfaced roads.
On this GM design each front surge tank was mounted
neatly inside the frame side-member, as shown in Fig. 8.10.
The bellows unit became a compact design of rolling
diaphragm (see Fig. 8.10). This diaphragm had a 3-in
diameter hole in the centre, thus making the volume below
the diaphragm part of the effective volume.
The rear axle was located by a four-link system and
suspended on an air spring at each side. Domed pistons or
pedestals, shown in cross section in Fig. 8.11, gave a rising
rate suspension in bump and an almost constant rate in
rebound. With more space available a large surge tank could
be provided at the rear.
The system was designed to operate with an air pressure of
75 Ibf/in 2 (5.17 bar). Spring frequencies were approximately
CONTROL YOKE AIRSPRIN5 WARNING LIGHT
PRESSURE SWITCH
CONTROL i.INK ACCU",UJ.ATOR TANK
AIR PRESSURE i./IVE
AIR EXHAUST LINE
Fig. 8.12 Layout of pneumatic suspension on Cadillac Eldorado.

The General Motors air suspension 141
0.9 Hz, or about 10% lower than the later models with coil
springs. As on the earlier bus design an air compressor, plus
an automatic levelling system, was provided.
8.4.1 Compressor and accumulator

An electrically driven piston-type air compressor with a
delivery of 600 in 3/min (0.59 m 3/h) was mounted above the
generator in the engine compartment. The compressor
operated upon demand from a pressure switch on the 500 in 3
(8200 ml) accumulator. This vessel also served as a trap for oil
or condensed water and was provided with a manually
operated drain valve. A warning light on the instrument
panel informed the driver if the accumulator pressure fell
below 70 Ibf/in 2 •
8.4.2 Control system

The complete installation is shown schematically in Fig. 8.12.
The Delco Products Division were responsible for the design
of the automatic levelling system. Besides a new lighter
levelling valve, they developed a system of solenoid valves to
allow the passage of air to the levelling valves when the
ignition was switched on or when a car door was opened.
With closed doors and the ignition off the solenoid valves
isolated the levelling valves. A second pair of solenoid
controls were used to allow a restricted air flow to pass in
and out of the air springs when the vehicle began to move.
These solenoids operated only on demand from the levelling
valves. Such action would be required if, for example, the
vehicle had been parked on a steep hill or with one wheel in
a pothole.
As in the earlier system designed for the Greyhound buses,
a delay mechanism was built into the levelling valves. No air
flow into or out of an air spring could take place under the
action of a movement that had a frequency greater than 10
cycles per minute (0.167 Hz). This ensured that each air
spring functioned as an individual unit under the action of
normal bump and rebound movements. Three levelling
valves were used, as in the bus installation, a Single valve at
the front and two at the rear.
References
[1] Sainsbury, J. H. (1958), 'Air suspension for road vehicles', Proc.

Inst. Mech. Engrs.
[2] Air Suspension and Servocontrolled Isolation Systems, Shock and
Vibration Handbook No.3, McGraw-Hill, New York, chapter
33, p. 331.
9
Hydropneumatic Suspension
9.1 The Citroen tradition
Engineers at Citroen have always been encouraged to be bold

in their thinking, even though the commercial history of the
company suggests that boldness can sometimes lead to
financial instability. Andre Citroen was the first automobile
manufacturer to make front wheel drive cars in large-scale
production. All Citroens since 1934 have been traction avant
and it is inconceivable that they will ever be otherwise. Even
though Citroen did fail financially in the 1930s, when an
injection of money from the Michelin Co. kept them in
business, the bold approach to engineering design was not
abandoned.
While Firestone and General Motors were wrestling with
the problems of pneumatic suspension in the 1950s, Citroen
forged ahead with their plans for a self-levelling suspension
that used a gas as the suspension medium and a liquid to
achieve the automatic levelling.
9.2 The Citroen suspension
Hydropneumatic suspension was first used in 1953 for the

rear suspension only, on the 6-cylinder Citroens. In 1955 it
was introduced on the OS model. The OS series of Citroens
was replaced by the CX series, a remarkable vehicle, using
high-pressure hydraulic circuits to operate not only a
self-levelling suspension system, but the brake system, an
assisted steering gear and an automatic control of clutch and
gear change. In this book our interest is in the suspension
only, but the reader must appreciate that the provision of an
expensive hydraulic pump and its ancillary controls is a
shared service. For obvious reasons provision is made in the
main hydraulic circuit, in the shape of a 'priority valve', to
144 Hydropneumatic suspension
isolate the suspension system and maintain normal hydraulic
supply and pressure to the brakes and steering in the event of
serious failure in the suspension system.
9.2.1 The pressure circuit

The heart of the system is the hydraulic pump, and to extend
the anthropomorphic analogy, we can see from Fig. 9.1 that
the hydraulic circuit contains arteries to supply oil under
pressure and veins or return pipes to return the oil to the
pump inlet. From this point our analogy begins to break
down. The Citroen engineers, being only human, found it
necessary to add several control devices that appear clumsy
when compared with the self-regulating devices found in the
human heart.
RH front suspension Brake pedal gear
RH front suspension
cylinder and sphere
R H rear suspension
cylinder and sphere
LH rear suspension
,,====~CY~lin;d~er and sphere
Rearheight
corrector
Gear selector
return
~-----~--
Pressure /Steering
regulator return
return
Fig. 9.1 Suspension hydraulic circuit.

The Citroen suspension 145
Suchon
Fig. 9.2 Simplified cross section of swashplate multicylinder pump

as used on ex model.
The Citroen pump on the larger models, the DS and CX

series, is an engine-driven, 7-cylinder swashplate pump,
shown in a simplified section in Fig. 9.2. It operates at
half-engine speed. On the GS model it is a single cylinder
unit. The pump delivery on the larger cars is 2.88 ml per
revolution. Since September 1966 (December 1968 in the
USA and Canada) a mineral oil, similar to engine-lubricating
oil, has been used as the hydraulic fluid. A spring-controlled
regulator is used to maintain a pressure in the main hydraulic
Nitrogen
ot the initiol
inflation
pressure
Nitrogen ot
the pressure
of fluid in use
Fig. 9.3 Hydraulic system fluid accumulator.

circuit within the designed limits. The regulator is set to cut
in at 140-147 bar (2030-2130 Ibf/in 2) and cut out at 162-167
bar (2350-2420Ibf/in2). These limits are wide enough to
permit a reserve of fluid to be stored in a gas 'spring-loaded'
accumulator. This reduces the number of starts and stops
demanded from the pump on a journey. Cross sections of the
accumulator are given in Fig. 9.3. The view on the left is of
the vessel in the original condition with the zone below the
Fluid from Cyl inder

the height ---->ooFH'-~~~
corrector
VJ:§:lE rrt-- - -- Piston -----I~~n
overfIOW_~===1
return ... Push rod
Venrto
atmosphere
Front
Rear
Disk tance
volves Calibrated spacer
bypass
Fig. 9.4 Front and rear suspension cylinders with enlarged cross
section of damper valve body.
synthetic rubber diaphragm at the initial charge pressure. The
gas is dry nitrogen. On the right the accumulator is fully
charged with hydraulic fluid at the regulator cutout pressure.
The pump is supplied with fluid from the fluid reservoir
which is not only a reservoir, but a filtration and degassing
tank into which all fluid is returned from the various
operating units, such as the suspension units, gear selector,
etc.
9.2.2 The suspension circuit

The sprung mass is supported on four units of the type
shown in cross section in Fig. 9.4. These are the type of unit
used on the medium-sized Citroen, the GS model. The upper
part of the 'sph~re' is charged with nitrogen (azote in French);
the lower part and the damper cylinder contains hydraulic
fluid. The gas cylinder is obviously not spherical, having a
cylindrical central zone. This profile is carefully designed by
Citroen to give the desired increase in effective piston area as
the nitrogen is compressed. This change in effective piston
area is illustrated schematically in Fig. 9.5. In the upper
diagram a wheel has risen in bump and the piston area is
almost the full 'sphere' diameter. In the lower diagram the
gas has expanded as the wheel falls into a pothole and the
diaphragm, having forced fluid from the cylinder, has now a
much-reduced effective piston area. This change in area with
diaphragm movement should be compared with the use of a
(a)
Fig. 9.5 (a) Compression of the gas spring under Single-wheel

bump. (b) Expansion of the gas spring under rebound.
tapered or curved pedestal to achieve a similar effect in the
Cadillac pneumatic suspension (see Fig. 8.8).
The movement of hydraulic fluid into and out of the lower
part of the 'sphere' is controlled by a damper plate in the
base. The damping action is double-acting. Small wheel
movements are absorbed by flow through the calibrated
bypass hole. Larger flows in both directions are controlled by
upper and lower disk valves. In this respect, the dampers
follow conventional damper practice.
9.2.3 Gas pressure

The load on each wheel is carried by the gas above the
diaphragm. The initial pressurization of each unit is related
directly to the weight carried. On the GS saloon, for example,
with a weight distribution of 62/38, each front suspension
unit is pressurized to 55 bar, the rear to 35 bar. In the
absence of wheel movement the pressures above and below
the diaphragms balance out to the same value. In the static
position each diaphragm will, therefore, be flexed to some
position similar to that illustrated in Fig. 9.4.
9.2.4 Height correction

Two height correction valves are used, one at the front and
one at the rear. These feed or extract fluid from the
suspension units to maintain a constant ground clearance.
Relative rotation of the anti-roll bar is used as an indication of
the correct ride height. The height corrector valve is shown in
section in Fig. 9.6. There are four pipe connections to the
valve. Three are shown in the upper cross section, these
being an inlet from the hydraulic pressure supply and a
connection to the reservoir at the top, an outlet to the
suspension units at the base. The lower view in Fig. 9.6 is a
half-section of the valve, but on a different plane passing
through the restricted passage (also called a 'dashpot' in the
Citroen literature) and the large-bore passage connecting
chamber 0 with chamber C. The action of the valve is
explained in the text included with Fig. 9.7.
Source of high pressure To reservo ir
Rubber
_~;...,..~~~---t:l-l,L..~_?" diaphragms
Disks ~====~;;"""I;:S:l::;;;::J~~
Springs ..",;::::::::,,-~s-1/T
Suspension cylinders
~~_f;jff..;j~- Restricted passage
Slide valve at central cut-off position

Fig. 9.6 Height corrector: a distributor block (3-way tap) which,
depending on the position of the slide valve connects the services
(suspension cylinders) to the inlet (HP supply); connects the services
(suspension cylinders) to the outlet (reservoir); isolates both the inlet
and outlet from the services (slide valve central). The chambers C
and D, sealed by rubber diaphragms (reinforced by metal cups), are
full of hydraulic fluid which comes from the seepage past the slide
valve; a seepage return takes the surplus fluid back to the reservoir.
Chambers C and D are interconnected by a clear passage drilled in
the sleeve of the slide valve, closed at each end by disk valves
controlled by the movement of the slide valve. In the central position
each disk is held against a face on the sleeve by a weak spring. A
restricted passage inserted in the body of the corrector (dash pot)
which limits the flow of fluid from C to D and back is connected to
the overflow return to the reservoir.
( b)
(d )
Fig. 9.7 Operation of the height corrector. (a), movement from
cutoff to exhaust position: When the slide valve is moved, i.e. when it
moves its position from cutoff, the disk valve in chamber C is held
on its seating by a return spring, thus closing the clear passage. The
disk valve in chamber D is lifted off its seating by the shoulder on
the slide valve, thus opening the free passage. The fluid in chamber
C is therefore obliged to pass through the dashpot which slows
down the fluid movement, which in turn slows down the movement
of the slide valve. Thus the slide valve will not move to the exhaust
position unless there is a positive effort on it for a certain period of
time. No correction occurs for rapid wheel movements; (b),
movement from exhaust to cutoff position: When the slide valve is
returned to the cutoff position, the fluid in chamber D can this time
use the clear passage and return to chamber C, lifting the disk valve
against its return spring. Return is therefore rapid. As soon as the
slide valve returns to the cutoff position, the disk valve in chamber
D closes the passage again, stopping the slide valve over-running the
cutoff position and avoiding a second correction; (c), movement from
cutoff to inlet position: When the slide valve is moved, the disk
valve in chamber D is held against its seating by its spring, closing
9.2.5 The function of the anti-roll bar in height control
Rotation of the anti-roll bar is used to control the movement
of the slide valve. We must first consider how rotation of the
anti-roll bar occurs. During roll one arm of the anti-roll bar is
forced upwards, the other arm downwards. Since these
movements are equal and opposite the bar is placed under
torsion, but no actual rotation occurs relative to the
supporting bushes carried by the body. Simple single wheel
movement in bump or rebound will produce some torsion
and a little rotation. However, if both front wheels or both
rear wheels move upwards or downwards in unison, the
anti-roll bar arms on both sides move in the same direction.
This produces rotation of the anti-roll bar and is the basis of
the Citroen system of height control.
The complete control mechanism is shown in Fig. 9.S. The
height control rod is clamped to the anti-roll bar and passes,
parallel to the bar, to a lever behind the corrector valve. This
lever is pivoted at its lower extremity. Rotation of the anti-roll
"bar moves the corrector valve slide inwards or outwards, as
described in Fig. 9.7. Rapid wheel movements, i.e. under the
action of single- or double-wheel bump or rebound, have no
effect on the corrector valve, since the movement of fluid
between chambers C and 0 is restricted by the 'restricted
passage' shown in Fig. 9.6. Height correction can only occur
after a delay of several seconds.
An over-riding manual control, located near the driver,
allows him to increase the ride height when travelling over a
rough terrain or to clear snowdrifts. It is also used when
changing a wheel.
the clear passage. The disk valve in chamber C is lifted off its seating
by the shoulder on the slide valve, thus opening the clear passage.
Liquid in chamber D therefore has to pass through the dashpot. As
for operation (a), there will be no movement of the slid~ valve until
a certain designed time period has occurred; (d), movement from
inlet to cutoff position: When the slide valve is returned to the cutoff
position, the fluid in chamber C can this time use the clear passage
and return to chamber D, lifting the disk valve off its seating. As in
operation (b), the return is rapid, As soon as the cutoff position is
reached, the disk valve in chamber C reseats. This stops any
over-run of the slide valve and prevents a second correction.
Suspension cylinder _ _~....-
Front 'halfaxle'
mounted on chassis
Anti - roll bar
Height corrector
~ Increasing load
[::>i> Decreasing load
Overflow returns
Control rod
Pivot point
Fig, 9.8 Activation of the height corrector valve by rotation of the

anti-roll bar.
The Citroen control system is excellent, for a mechanical

system, but there is little doubt that a much lighter and
cheaper electronic control system will be seen in the
not-too-distant future
9.2.6 The GS front suspension

On the GS model a double-wishbone layout is used for the
front suspension with the spring unit acting on the upper
~;>.--_ _ Suspension
{,.--_-li\ sphere
Suspension cylinder
Bump as rebound
fers
Height
corrector
Fig. 9.9 Front suspension on GS model.
wishbone. The piston stroke of 70 nun becomes a total

movement of 210 nun at the wheel. The mechanical
advantage can be seen in Fig. 9.9. With a self-levelling
system this total \yheel movement of 210 nun is available
even when the vehicle is fully laden. The spring rates for
small wheel movements are exceptionally low. Measured at the
wheel they are approximately 5 kN/m one-up rising to 6 kN/m
fully laden. The natural frequencies given by these two rates
are 0.65 Hz and 0.70 Hz respectively. These frequencies are
nearly one-half the values given by coil springs on
conventional vehicles. The spring rates in bump rise rapidly
as the deflection increases. Not only does the gas pressure
rise under compression, but the diaphragm area increases. In
re bound the spring rates and frequencies are still very low for
small 'wheel movements. In this direction, however, the gas
pressure falls as the deflection increases and the diaphragm
area also becomes much smaller near the limits of rebound
movement. The flow characteristics of the damper valves also
influence the effective spring rates. By providing stronger
springs on the rebound damper valves, Citroen are able to
provide an adequate control on wheel movement in rebound.
Rubber buffers, indicated as item 1 in Fig. 9.9, act as bump
and rebound stops.
9.2.7 The GS rear suspension

A trailing link rear suspension is used with the suspension
unit set at an angle of 17° to the horizontal. The general
layout can be seen in Fig. 9.10. The suspension unit is
hidden by the subframe in this view; only the top of the
cylinder is visible. The details of the suspension cylinder and
sphere can be seen in Fig. 9.11. For small wheel movements
the frequency in bump and rebound at the rear is about 15%
higher than that at the front.
9.2.8 The anti-roll bars

The rear anti-roll bar is 18 mm diameter, the front 21 mm.
Since torsional stiffness varies as the fourth power of the
diameter, the front anti-roll bar rate is 85% greater than the
rear. Even though the centroid of the sprung mass has a
strong bias towards the front, there will still be a greater roll
Flexible Flex ible Flexible

mounting moun~jng Fuel ~ank mounhng
Suspension Wheel bar

cyl inder
Fig. 9.10 Rear suspension installation on GS model, showing

anti-roll bar.
Ant i-roll
bar
Fig. 9.11 Rear suspension cylinder on GS model and its installation

details.
resistance at the front and a slight tendency to roll-oversteer

when driven near the limit in a corner.
The Cadillac engineers did not use anti-rolls bars in their
self-levelling system, and with a delay mechanism built into
the height control mechanism there would be a tendency to
roll excessively when making rapid changes in direction. This
pro blem does not arise with the Citroen design.
9.2.9 Ride-height adjustment

I have seen how useful this device can be. I lived at one time
within a kilometre of the west shoreline of Lake Michigan,
seeing my doctor neighbour in his DS model Citroen charge
quite successfully through formidable snowdrifts. On such
mornings he jacked up the Citroen to its full 254 mm (10 in)
ground clearance. His most reliable Volkswagen Beetle was
used only after the city snowploughs had cleared the streets.
9.2.10 Cornering power

Only in one sense does the Citroen appear old-fashioned.
Trailing link suspension is used at the rear and this, we
know, gives undesirable camber changes in roll. At the front,
although the double-wishbones are angled inwards towards
the car centre line, they are almost of equal length and will
give an undesirable positive camber to the outer wheel under
roll. The Citroen, therefore, does not appear to achieve the
full potential of the cornering force available in the tyres.
Other suspension geometries are possible and offer greater
cornering potential. This will be seen when we examine the
latest European application of hydro pneumatic in the S-class
Mercedes-Benz 450 SEL later in the chapter.
9.2.11 Anti-squat
It would be very odd indeed if Citroen introduced all this
complex hydraulics into the suspension design in order to
provide a constant ride height and then completely ignored
the longitudinal dips and dives induced by braking and
acceleration.
Anti-dive geometry in general was discussed in Chapter 4.
)F $
The centroid in the Citroen is well forward, with a one-up
weight distribution of 62/38 and a full-load distribution of
$£':-: ___= __
Centroid
(T (f"";'OdOn)
, - -$,;:... I~
Centroid
(one-up) (fully laden)
Of Fi=r2 Pr
,~-~J
p. H
-- 1I==-'--+-
f
Braking
Fig. 9.12 Anti-squat and anti-dive geometry on GS model.
54/46. The force diagram in Fig. 9.12 is a compromise based
on a weight distribution of 58/42.
To achieve perfect anti-squat geometry, the decreased loads
on the front footprints and the increased loads at the rear, as
shown in the upper diagram in Fig. 9.12, must be balanced
by a vertical downthrust on the suspension arms at the front.
The tractive force F from the front wheels produces an
acceleration IX. Approximately 10% of this tractive force is
expended in accelerating the unsprung mass and about 90%
in accelerating the sprung mass M s:
0.9F = F j = MslX.
F j is the inertia force acting on the sprung mass. This tends to
rotate the sprung mass about its suspension anchorages. This
decreases the load on the front wheels by AL and increases
the rear wheel loads by the same amount:
H H
AL=F.-=MIX-
IE s E .
The effective arms of the two moments Hand E can be seen

in Fig. 9.12.
To balance this tendency for the body to rotate during
acceleration, the front wishbones have been angled upwards
on the Citroen GS at an effective leading arm angle of 12
degrees. Taking Fa = F j = 0.9F, where F is the total tractive
force, f = Fa tan 12° .
From Fig. 9.12 it is seen that perfect anti-squat geometry is
achieved when the resultant force F r is angled towards the
centroid. The diagram shows that almost perfect anti-squat
balance is given when fully laden. With a one-up load about
80% anti-squat is given.
9.2.12 Anti-dive
With a trailing arm suspension at the rear and a wishbone
system at the front, angled to give an effective leading arm
action, the Citroen GS has an ideal geometry for anti-dive
correction. As shown in the lower diagram in Fig. 9.12, the
values of AL now become an increase in load at the front and
a decrease at the rear. The braking forces applied to the
footprints B f at the front and Brat the rear, produce reactions
at the suspension points P f upwards at the front and P r
downwards at the rear. These forces, with careful design,
will balance the couple AL x E.
The Citroen GS has a brake pad area at the front of
146 cm 2 and at the rear of 68 cm 2 • This suggests a front/rear
braking effort of 68/32. The situation is more complicated,
however, since the front and rear brake cylinders do not
operate at the same pressures or even at constant pressures.
The front brakes always operate at normal full-circuit
hydraulic pressure (circa 160 bar). The rear brakes operate at
rear suspension pressure, which is nominally 35 bar. When
fully laden the rear suspension pressure increases to as much
as 65 bar, but this is still well below the operating pressure at
the front. It is obvious that the front braking bias is so high
under all load conditions that the front wheels will always be
the first to lock. This is an established prerequisite for safe
braking in the wet. The lower diagram in Fig. 9.12 is given
purely to show the principles of anti-dive geometry, using
leading and trailing arms. We do not have sufficient data to
put concrete values to the Citroen design.
9.3 The Mercedes-Benz hydropneumatic suspension
The engineer has always been in control at Daimler-Benz.

The German for 'stylist' is no doubt a very long word, but
his influence is relatively short. Market forces are not ig-
nored, but a new body on a Mercedes-Benz is normally only
introduced as outer covering for a new advanced design of
car. The engineering comes first, the styling second.
Having abandoned their long battle with the swing axle
rear suspension, Daimler-Benz adopted the semi-trailing arm
rear suspension and refined it. In conjunction with a
well-developed double-wishbone system at the front, using
unequal-length wishbones suitably angled to give anti-dive
geometry, the mid-1970s range of Mercedes-Benz cars
seemed to represent the ultimate in conventional suspensions.
When tested by Motor in 1974 the 450 SEL was pronounced
as 'simply magnificent'. Even while the Motor staff were
revelling in the 'uncannily roll-free, sure-footed handling', the
The Mercedes-Benz hydropneumatic suspension 159
research-and-development engineers at Stuttgart were
working on a hydropneumatic suspension.
9.3.1 The 450 SEL 6.9

The 1979 450 SEL is powered by a 6.9 litre V8 engine of
210 kW (286 h.p. DIN), with K-Jetronic petrol injection. The
most interesting feature, however, is the application of gas
suspension with hydraulic level control to the well-
established mechanical suspension components from the
earlier S-series Mercedes-Benz chassis.
Interest in the Experimental Department in hydropneumatic
suspension began in the 1950s, obviously inspired by the
Citroen DS 18. Daimler-Benz made a serious study at the
time of the influence on roll and pitching of the use of
gas springs. The first application of this work was
a hydropneumatic shock-absorbing strut used as a
compensating spring in conjunction with the low-pivot,
swing axle suspension used on later models of the 300 SL
sports car. The next development was a self-levelling rear
suspension using hydropneumatic struts with pressure
supplied by an engine-driven pump. This, in tum,
progressed to a fully automatic hydropneumatic suspension a
la Citroen. This is available on the S-class top model, the 450
SEL 6.9.
There are improvements on the Citroen system. For
example, suspension height can change when the engine is
not running. This is because the central reservoir is charged
to a higher pressure than that required for normal operation.
This high-pressure reservoir is particularly valuable when
changing a flat tyre. The pressure line to any particular
suspension unit can be isolated during a wheel change then
repressurized from the reservoir afterwards. If the pressure
supply fails by pump failure or other cause, both axle circuits
can be isolated to complete the journey. Another
improvement is the basic suspension geometry which, unlike
that on the Citroen, is designed to maintain near-verticality of
the outer wheels when cornering.
The layout of the hydraulic system is shown in Fig. 9.13
and the placing of the suspension components in Fig. 9.14.
One clever modification of the Citroen system is the use of a
Leak oil return line
sure line
Control pressure line
Fig. 9.13 Hydraulic circuit for hydropneumatic suspension on Mercedes-Benz 450 SEL 6.9.
c::
-
o
'"c::
~
8-
E
o
u
Damper pi~ton
Pressure
Piston rod ----H;;.. line
Pressure cylinder
Fig. 9.15 Front wheel suspension element on Mercedes-Benz.
large-bore connecting pipe between the hydraulic suspension

cylinder and the gas 'sphere'. This gives the designer more
freedom in the disposition of the components. Fig. 9.16
shows how the gas reservoir (the sphere) can be fitted
neatly into the side of the rear wheel arch. The Citroen
engineers solve the difficult problem of intrusion into the
passenger space by placing the cylinder/sphere unit in a
near-horizontal position. Fig. 9.15 shows a cross section of
the Mercedes-Benz front suspension cylinder, and Fig. 9.16
shows the rear suspension layout. It will be seen that a
Suspension element
with shock absorber
Gas-filled
pressure reservoir
Fig. 9.16 Rear wheel suspension layout on Mercedes-Benz: 1,

pressure line; 2, gas-filled pressure reservoir; 3, suspension element
with shock absorber; 4, torsion bar; 5, level control; 6, diagonal
control arm.
variant of the Citroen level control method is used. Angular
movement of the centre point of the anti-roll bar (see Fig.
9.14 for the position of level control clamp) is used to actuate
the level controller (item 5 in Fig. 9.16).
The detail design of the suspension cylinder can be seen in
Fig. 9.15. The damper valves are carried in the piston, not in
the base of the gas reservoir as in the Citroen. Any leakage
past the piston rod seals is returned to the main reservoir via
the special return lines. The hydraulic supply reservoir is the
large cylindrical vessel which can be seen in Fig. 9.13 situated
in front of the front suspension cylinder on the right. The gas
vessel associated with this reservoir can be seen adjacent to
the suspension 'sphere' on the right of the drawing.
Anti-dive and anti-squat had already been built into the
450 SEL suspension before the adoption of hydropneumatic
suspension. Anti-roll provision, however, had to be
supplemented. With gas suspension bump and rebound rates
very much lower than those used with coil suspension, it was
necessary to provide much stiffer anti-roll bars to achieve the
same roll resistance.
10
Interconnected and No-roll
Suspensions
10.1 Interconnected front/rear suspension
The concept of interconnecting the front and rear suspensions

to reduce the period in pitch is an old one. A mechanical
interconnect was developed for the little Citroen 2CV and
this, despite an obvious lack of roll stiffness, gave a very good
ride over rough terrain for such a tiny vehicle.
Let us consider as a first step a simple system where both
front and rear suspensions share the same spring, i.e. one
spring per side as shown in Fig. 10.1. With a perfectly
symmetrical system the sprung mass could not oscillate in
pitch, since a downforce at A would deflect the front spring
connection a distance x, and this would be balanced by an
upforce at the rear of identical value which would deflect the
rear spring connection an equal distance x in the same direction.
Spring deflection would therefore only occur in bump and
rebound, not in pitch. Before we rush off to the Patent Office,
let us consider what would happen to such a vehicle on the
road. As soon as we started to move from a standstill, even at
a negligible rate of acceleration, the inertia force of the sprung
F F
Fig. 10.1 An interconnected suspension, but with no resistance to

pitch.
166 Interconnected and no-roll suspensions
mass would rock the body backwards, pushing downwards
at B and pulling upwards at A. There would be no spring
force to act against this inertia force and the body would tip
backwards hard against the limit stops. Light braking would
produce the opposite reaction, since the system is completely
unstable in pitch.
10.2 The Citroen 2CV
Additional springs without interconnection are seen to be

essential to achieve a practical interconnected suspension.
This can be seen on the Citroen 2CV, shown schematically in
Fig. 10.2 in which the two main springs 5 f and 5 rare
enclosed in a 'floating' cylinder, which is mounted between
auxiliary springs Sf and s r with abutments against the sprung
mass. In the later examples of the 2CV these auxiliary springs
are made of rubber.
Considering a simple system with a weight distribution of
50/50 and identical springs rates front and rear (5 f = 5 rand
Sf = Sr), we have a system in which bump and rebound is
controlled by two springs 5 and s working in series and pitch
is controlled by two springs 5 and s working in opposition.
When two springs of rates 5 and s work in series, they act as
if replaced by a single equivalent spring of rate 5 e where:
5 = 5 xs
e 5 +s .
Sf Sf Sf Sf
Fig. 10.2 An interconnected suspension as used on 2ev Citroen.

The Moulton hydrolastic suspension 167
If, for example, we make s = 0.75:
5 = 5 x 0.75
e 5 + 0.75
= 0.4125.
5 P' the rate in pitch = 5 - s
= 0.35.
This would give a natural frequency in pitch (since natural
frequencies vary as the square root of spring rates) of about
85% of the natural frequency in bump and rebound. The
above analysis is an oversimplification, but serves to illustrate
the Citroen principle.
10.3 The Moulton hydrolastic suspension
A mechanism using hydraulic interconnection of front and

rear suspensions was filed with the British Patent Office in
Fig. 10.3 Idealized behaviour of the MG 1100, with interconnected

suspension units, when traversing a bump in the road.
Damper
valve
Tapered
cylinder
j)iaphra~m
NORMAL BUMP
Rubber spr;n~
Damper hleed-.-;;="'="nll
Tapered piston
REBOUNLJ (Sect"ioned at"

90° to other vlews)
Fig. 10.4 The BMC Hydrolastic suspension unit.
1955 by Alex Moulton and tried out experimentally in the

same year on a prototype Alvis that had been designed by
Alec Issigonis.
The first production application of Hydrolastic suspension,
the name adopted by M.oulton Developments Ltd, was to
the BMC 1100. An idealized diagram of the behaviour of
hydrolastic suspension on a variant of the original ADO 17
design, the MG 1100, is shown in Fig. 10.3. The suspension
unit is shown in section in Fig. 10.4. It consists of a tapered
The Moulton hydrolastic suspension 169
sheet-metal casing, fixed to the body frame, which encloses in
the upper section a conical rubber spring. At the base is a
sheet-metal partition. The outer rim of this and the lower
extremity of the casing are swagged around the reinforced rim
of the moulded rubber/nylon reinforced diaphragm. Two
rubber one-way valves in the centre of the partition permit
flow in both directions. These two valves, together with the
small bleed hole, replace the conventional damper system.
The fluid on both sides of the partition is a mixture of water,
alcohol and anti-corrosion agent. The piston rod which is
located in the centre of the diaphragm is connected to the
suspension links and moves up and down with wheel
movements.
10.3.1 Interconnection
The Hydrolastic spring elements on the same side of the car
are interconnected by small-bore piping. The pipe connection
'cannot be seen in Fig. 10.4 but is made at a high point on
the upper surface. Fig. 10.3 illustrates how single wheel
movements were controlled by both front and rear springs on
the same side working in series. With very large-bore piping
this would have effectively halved the spring rate in
comparison with the use of isolated units. In practice the use
of small-bore piping modified the true series operation to an
intermediate effective spring rate. The damper design also
modified the overall behaviour. Under roll, however, the two
springs on the same side worked in parallel and, as we
know, springs working in parallel are additive.
The Moulton Co. have a long experience of the design of
rubber suspension units and they were able to provide a
rubber doughnutlike spring (see Fig. 10.3) which had been
carefully designed to give a rising rate suspension. With very
large-bore pipes the only resistance to movement in pitch
would be the resistance to flow designed into the damper
valves. With small-bore pipes the pitch frequency must
increase for larger movements. From experience, I can state
that pitch control on this vehicle was very good and for such
a small car the ride was excellent. Today some small cars with
conventional springing have an inferior ride.
- .
HYDRAGAS
...
//~~"\16_ Lorge Small
area area
PI TCH
Gas
HYDRAGAS
BOUNCE
Fig. 10.5 The Hydragas principle.
10.4 The Moulton Hydragas suspension
The success of the Citroen hydropneumatic suspension no

doubt inspired the Moulton Co. to tum to gas as the
suspension medium; also to use a tapered pedestal or piston
to give a variable effective diaphragm area, the technique
originally used by Firestone.
The Hydragas principle is shown schematically in Fig. 10.5,
and the detail design of a Hydragas spring unit is shown in
Fig. 10.6. It will be seen that the effective piston area acting
on the diaphragm increases with increase of wheel movement
in bump and decreases under rebound. Nitrogen gas under
pressure and compressed by a flexible diaphragm acts as the
spring. Even in the position of the damper valve, the Moulton
gas spring is almost identical to the one in the Citroen
system. In the Hydrolastic suspension system the
interconnection between front and rear was made from the
liquid compartment above the damper valves. In the Hydragas
system the connecting pipes are below. This important
difference means that liquid flow between the front and rear
units does not pass through the damper valves in the latest
system. The main resistance to pitch is given by the change in
diaphragm area by the piston taper, as shown in Fig. 10.5.
Since one diaphragm increases in effective area as the other
decreases the system behaves in pitch like a conventionally
sprung car, but with no conventional damping. Some
damping is given by the long interconnecting pipes.
Rates. The makeup of the bounce, roll and pitch rates, using
the terms used by Moulton Developments Ltd, is as follows:
Hydraulics* - that due to the compression of the gas acting
through the liquid (at the same pressure) upon the
diaphragm.
Taper - that due to the pressure of the liquid acting upon
the changing area of the diaphragm, as it is actuated by the
suspension arm.
Parasitic - that due to rubber bushings in the complete
suspension system.
Drop angle - that due to change of leverage with stroke.
Essentially in the Hydragas system the pitch rate is much
lower than in bounce; the bounce rate and the roll rate are
identical since no anti-roll bars are used. As shown in Fig.
10.7, the pitch rate is made up of taper, parasitic and drop
Rear
Fig. 10.7 Makeup of bounce, roll and pitch rates in Hydragas

system.
• Why this is not called 'Gas' or 'Pneumatic' rate is not clear.

No-roll suspensions 173
angle. The bounce and roll rates include these three plus the
hydraulic rate.
On a typical Hydragas suspension such as the Austin
Allegro the pitch rate would be 13 kN/m, the front bounce
and roll rate, 20 kN/m and the rear bounce and roll rate,
18 kN/m. On the 1750 HL Allegro, with a fully laden sprung
mass of about 1050 kg and a front/rear weight distribution of
63/37, the front period in bounce is approximately 1.2 Hz and
the rear 1.5 Hz. The pitch frequency is about 1.0 Hz, a very
low value and well removed from the bounce frequencies. I
myself have used an Allegro for personal transport for several
months and found the ride to be as good as that given by
many larger cars. The low pitch frequency is no doubt
responsible for the favourable comments on the good ride
given in the back seat.
10.5 No-roll suspensions
10.5.1 Roll, no-roll or banking

Before wasting time and money on making a product, we
should always establish that a need really does exist. The
Ford Co. no doubt still remembers the Edsel. There was
nothing much wrong with the Edsel, but the American public
was not in the market for yet another big car when they
introduced it!
Very little market research seems to have been carried out
to discover how much the typical carowner will pay for a car
with a better ride, or a reduced tendency to roll when
cornering. Even the men who design the cars do not seem to
be of one mind when the question of roll is considered.
French cars, for example, usually roll more than British cars.
What, indeed, is the ultimate suspension behaviour when
cornering? Should there be a limited degree of roll, no roll, or
even what is usually called 'banking', i.e. a reverse body roll
to counteract the effects of centrifugal force on the occupants?
10.5.2 The banking car

The idea of banking the body, i.e. leaning inwards, when
cornering has appealed to many inventors. It appealed
especially to the designers of racing motor cycles fitted with
sidecars. A banking sidecar seemed a more sensible approach
than one where the sidecar passenger was hanging perilously
from one side or the other. Even so the idea of the banking
sidecar eventually died.
Against this failure we can see a successful application in
British Rail's high-speed train, which has a mechanism to
make the carriages bank up to an angle of 9° when
negotiating curves. This is simply to reduce the side forces on
the passengers. A similar mechanism could also be used on
an automobile.
In 1961, I drove a Chevrolet Impala sedan that had been
converted by Mr J. Kolbe of Menomenee Falls, Wisconsin,
USA to bank when cornering. Measurements taken from a
photograph show the stock Impala with a body roll of about
6° on a particular turn. At the same speed on the same curve
the Kolbe-converted Impala banked to an angle of about
2°30'. The Kolbe Chevrolet was a very restful car to drive in,
travelling along winding country roads, since the full-width
front seat of the Impala gave very little lateral support to
driver or passengers. In Suffolk, where I now live, a banking
car would be a welcome innovation. In the USA it was
difficult to find suitable winding roads on which to test out
the Kolbe car. In general American roads travel in a straight
line for mile after mile. Eventually, a 90° turn is made on to
another long straight road. Little time is lost if the driver takes
these 90° turns at a fairly low speed. Sad to relate, but Mr
Kolbe invented the right car in the wrong country. The need
must always come before the invention.
10.5.3 Practical no-roll suspensions

The advantages of no-roll suspensions have been enhanced in
recent years by the increasing need to keep the wheels
vertical in a corner. This applies more to sports cars and
racing cars. For the typical tyres used on family saloons there
is little reduction in cornering power until positive camber
angles of 3-4° are exceeded.
Of course, it is possible to design front and rear
suspensions with roll centres at the same height as the
body-centroid centre line. With most vehicles this would
require a very high roll centre at both ends and high roll
centres result in large camber changes in· single wheel bump
and rebound, if we are to confine our design to conventional
practice. This was clearly demonstrated in Fig. 5.15 in
Chapter 5. With less popular suspensions such as the Morgan
and the older Porsche parallel trailing arms there is no choice,
since the roll centres are at ground level. The de Dion system
gives a high roll centre, and it was rumoured several years
ago that Team Lotus had considered the possibility of using a
de Dion axle at both ends.
All Formula 1 and many Formula 2 cars in 1979 were what
are popularly called 'ground-effect' cars. As was discussed in
Chapter 6 these cars are designed to create a Venturi-effect
underneath the car body to produce a surprising downforce
thus increasing the available cornering forces in a high-speed
corner by as much as 40% It is essential in such a system to
maintain a good air seal between the sides of the body and
the road surface. Sliding skirts have been developed for this
purpose, but two critical factors threaten their viability. The
first is the condition of the track surface, the second is the
extent of the body roll. Modern billiard-table surfaces on
Formula 1 circuits have made it possible for relatively short
skirts to be used, moving up and down in the slides under
bump and rebound and roll. Obviously, any tendency to roll
1550 Nm (40"/0)
~
1~8°
il
_.-tI
Centroid of sprung mass
4400 N
I 1600 N
j .- -
0, =--:-------~-~ ~O,
Roll centre
2935 N (28"/0) 1070 N (10"/0)
Fig. 10.8 Analysis of tyre cornering forces and roll moment at front
of Lotus 79.
00
I
2330 Nm (60%)
~
1f- 20
I Cen~roid
.1
of sprung mass
I II
'.
---~
I
I I
6450 N
~~~-.r--4~--~====~~_~_~~~~~~
il __ ~
Roll centre 0,
4670 N (45%) 1780 N (17%)
Fig. 10.9 Analysis of tyre cornering forces and roll moment at rear
of Lotus 79.
on comers as much as typical racing cars of the early 1970s

did calls for the provision of very cumbersome deep slide
mechanisms.
Modem low-profile racing cars have placed a tight limit on
roll for almost a decade. Even so, in 1978 when the Lotus 79
first appeared with sliding skirts, there were still a few cars
that rolled as much as 3° when cornering near the limit,
although the majority only reached about 2°, the limit for a
viable ground effect car. Figs. 10.8 and 10.9 are based on
measurements reported by Motor (March 3rd, 1979) in a
150 m.p.h. (67 m/s) comer and show the Lotus 79 rolling at a
value very close to this limit.
The disposition of the main masses, the engine,
transmission, radiators, fuel tanks and driver on a GP car are
as close to the ground as the designer can put them. In the
case of the typical 1979 Formula 1 car, the sprung mass
centroid centre line slopes slightly upwards from about
150 mm above ground level at the front to about 180 mm at
the rear. Since the roll centre is very close to ground level,
the roll moment in a comer taken near the limit is very high.
Extremely stiff springs are required to resist this moment.
Even though progressive rate springs are used, the ride is so
hard that young men in excellent physical condition feel
battered at the end of a race.
Figs. 10.8 and 10.9 show that the inner tyres of an F1 only
contribute about 30% of the total cornering force in a
150 m. p.h. comer. If then we could design a racing car with a
'no-roll' suspension in which load transfer did not take place,
we could make more efficient use of the inner tyres and
corner at even higher speeds. What is of equal importance
(until they ban the ground-effect carl) is that a 'no-roll'
suspension would assist the designer in the design of a
ground-effect body.
10.5.4 The Trebron suspension

Norbert Hamy is a Canadian architect who invented the
Trebron double-roll centre suspension more than a decade
ago and has progressed, with the inevitable blind-alley
developments and feasibility studies, to the latest Trebron
Concordia, built as a prototype sports car with the
collaboration of the Faculty of Engineering at Concordia
University.
The earliest feasibility study was carried out by Harry
Ferguson Research Ltd, resulting in the conversion of a Ford
Escort by Broadspeed Engineering for Group 5 racing.
Unfortunately, new racing regulations outlawed suspension
changes in this group before this conversion could be
properly sorted out on the race track but the initial
impressions were very promising.
Detail drawings of the Concordia project have not been
made available, but an earlier design for a Trebron Formula 1
chassis is given in Figs. 10.10 and 10.11. In this case the
designer has concentrated on maintaining verticality of both
outer and inner wheels. A camber angle of 0°-0°5' is no
mean achievement at a cornering acceleration of 1. 4g. The
secret ingredient of the Trebron system is the provision of a
separate bulkhead with a roll centre RC 2 high in the air.
Under the action of centrifugal acceleration the bulkhead rolls
inwards, i.e. it banks. The roll centre for the suspension
system RC 1 is at ground level. The banking action of the
bulkhead produces a rocking action in the rocking links by
which the upper wishbones are attached to the bulkhead. In
the case of Fig. 10.11 the bulkhead has not only rolled
inwards by 3°30', but shifted bodily sideways under
\
/
I
/
/
RCl
Ground
Fig. 10.10 Trebron double roll-centre suspension applied to racing car - static condition.
j t
'b
I
'"
.!l
'b 0
f"
ll=-~
Jill
Fig. 10.11 Trebron DRC racing chassis under cornering acceleration

of lAg. The susl'ension bulkhead swings about the upper roll centre
(RC 2) and the wheels remain upright.
centrifugal force by an appreciable amount relative to the
body (or chassis) centre line. The Trebron system shows great
promise and might eventually be the answer to the
contemporary racing car designer's most pressing problem.
10.6 The fully stabilized or 'active' system
Time and again, we have used the word 'compromise' in this

introduction to the technology of automotive suspensions.
There are fundamental limitations to conventional suspension
systems. One basic limitation is that the static deflection
varies as the inverse square of the natural frequency of the
spring. Attempts have been made to overcome this problem
by using rising rate springs, but this is only a palliative, and
many research workers in the field have been studying the
application of 'closed-loop' control systems to automotive
suspensions and there have already been several experimental
applications.
10.6.1 Active suspensions

A system is active as opposed to the conventional passive
system when it uses a fast-acting, closed-loop control system
of the type employed in aeronautics for many years. Such a
system uses hydraulic rams, servovalves and sensors. A
closed-loop control system requires the interconnection of
these elements to feed back signals from the hydraulic rams
by means of the sensors to activate the control device, which
in tum feeds signals back to the rams. The number of sensors
and the complexity of the control device depends upon the
number of variables we find it necessary to control. Studies
that have been made of the potential advantages of an active
suspension system suggest the following:
(a) a low natural frequency;
(b) low dynamic and static deflection;
(c) high speed response;
(d) no change in frequency or response with change in load.
The disadvantages are the obvious ones of increase"
complexity and increased cost.
The fully stabilized or 'active' system 181
10.6.2 Automotive products no-roll suspension
Automotive Products Ltd have developed an active
suspension system that is fully stabilized, i.e. pitch and roll
are virtually eliminated and vertical accelerations are
considerably reduced when compared with conventionally
sprung cars. The system is still under development at th~ir
Leamington Spa research establishment, but a prototype form
has been tested on a Rover 3500.
In the AP system the levelling valve no longer has a time
delay built into it. Every effort is made to eliminate this time
delay. If one stands on the front bumper of a Citroen CX
model (with the engine running), the front end dips, then
over a period of about two seconds, slowly rises to the
controlled height. If one stands on the front bumper of the
Rover 3500 equipped with the AP suspension system,
nothing appears to happen since the levelling valve reaction
time is a small fraction of a second. A line drawing of the AP
system is given in Fig. 10.12. Power to correct ride height is
provided by a pump capable of a very hi~h oil flow at a
pressure of 138-170 bar (2000-2500Ibf/in). It is a much
larger pump than that used by Citroen. Full pump output,
which would only be required occasionally as when cornering
at the limit, is 10 kW (13.4 h.p.).
Hydrauli c Main
strut Rea r roll
c ontrol
Gas sp ring
an d
control valve control valve damper
Fig. 10.12 Line drawing of AP fully stabilized suspension.
As in the Citroen system, each wheel is provided with a
gas spring and damper strut which is connected to a
suspension arm. The level of the car body under the action of
roll or pitch is corrected by very rapid additions or extractions
of hydraulic fluid from the space between the gas spring and
the damper piston. The signal to add or extract fluid is given
by a three-port valve. What the maker calls a 'pendulous
mass' is used to sense any change in level (see Fig. 10.13).
This mass is supported on a spring with a small hydraulic
damper in parallel. For any application the
mass-spring-damper unit must be specially tailored to match
the suspension system. In effect, it is a replica of the car's
suspension system in miniature. Thus under single wheel
bump the action is as follows: upward movement of the
suspension arm compresses the gas spring and creates an
upward force to lift the body. The pivot of the offset
pendulum also moves upwards, since it is attached to the
body. If the pendulum did not move, the spool of the
three-port valve would move to the left and extract fluid
from the suspension leg. The pendulum-spring-damper unit,
having been designed to behave as a replica of the
suspension system, will produce an upward movement of the
pendulous mass at an identical velocity to that of the body.
~~~~jjlsupporting
t~~~~~ spr ing
\:jg~~:::;;:;::~ Pendulous
pump
Cont rol va lve and suspension
Fig. 10.13 One suspension leg of the AP system, showing

schematically the hydraulic control system.
The fully stabilized or 'active' system 183
The net effect is that the valve spool will not move in bump
or rebound of a single wheel. The pendulum-spring-damper
unit therefore acts as a filter for transient signals from a single
unsprung mass, i.e. one wheel. Movement in roll, in brake
dive or in acceleration squat - unless they are very transient
as one might produce by a slight wiggle of the steering wheel
- are usually at a lower frequency than single wheel
movements. The valve spool reacts to these lower-frequency
movements and fluid is either pumped into the appropriate
struts or extracted from them at the combined dictates of the
valves. In the experimental Rover two control valves are used
at the front and a single control valve, with a lower response
time, at the rear. The two front valves control the body level
in pitch and roll, the rear controls only pitch. All three valves
also maintain a constant body height irrespective of load, as
in the Citroen and Mercedes hydropneumatic suspensions.
Early experimentation on the Rover installation showed that
diagonal connection of the rear struts with the front gave
good control of roll, but the rear struts had to be correctly
sized relative to the front to give the correct balance between
the front and rear roll couples. When the correct balance of
strut sizes had been achieved, the Rover could negotiate
bends and chicanes with precision and negligible roll. In fact
the only observable roll was produced by tyre compression.
When tested by the staff of Motor, the AP-stabilized Rover
could negotiate a test chicane at 54.6 m.p.h. The limit for a
standard Rover 3500 was 48.5 m.p.h.
The AP system would not be cheap. The pump alone
would cost about £100, if produced in modest quantities. If
one allows a total additional cost of £250, this would be a
negligible increase on a car costing £15000-£30000, such as
a Jaguar, BMW or Mercedes. In the price-conscious mass
market it would not sell as well, unless perhaps one also
included with the package a chromed embellishment in a
prominent position on the body to tell the Jones's that the car
was 'float-ride stabilized'.
References
[1] Moulton, A. E. and Best, A. (1979), 'From Hydrolastic to

Hydragas', Proc. Inst. Mech. Engrs, vol. 193, no. 9.
[2] Moulton, A. E. and Best, A. (1979), 'Hydragas suspension', SAE
paper 790374.
[3] Hegel, R. (1973), 'Vehicle attitude control methods', SAE paper
730166.
[4] Sutton, H. B. (1979), 'Synthesis and development of an
experimental active suspension', Automobile Engineer, vol. 4, no.
S (October-November).
11
A Small FWD Saloon Car:
Ford Fiesta S
11.1 Front wheel drive
Today everyone seems to accept FWD, usually with a

transverse engine and gearbox, as a sine qua non in the
smallest class of four-seater family saloon. No other design
gives as much space for the occupants and the luggage. FWD
is effective, inexpensive and reliable. It was not always so.
The first FWD car that I owned was a 1936 BSA Scout
with front drive shaft couplings that produced unpleasant
alternating reactions on the steering gear in a tight turn. Some
drivers chose to ignore this kickback at the steering wheel; it
is even possible that some regarded it as indication that they
were cornering at too high a speed. To an engineer it was an
indication of a variation in drive torque, something that
would soon produce excessive wear. It was no consolation to
me when my fears were proved right!
11.2 The constant-velocity universal joint
In this book we are not concerned with the transmission of

power to the road wheels, but the rapid expansion of FWD
automobiles since the end of the Second World War owes so
much to the invention of the constant-velocity joint that we
must consider in some detail the mechanics of this device.
Jerome Cardan, a sixteenth-century physicist, invented the
first mechanical joint for transmitting torque from one shaft to
another set at an angular displacement to the first. The first
practical universal joint, however, was devised by Robert
Hooke about 300 years ago and this well-known joint is
illustrated in Fig. 11.1. This very simple joint, in many
mechanical guises, was the basis of all mechanical U/J for
186 A small FWD saloon car: Ford Fiesta S
Fig. 11.1 Double Hooke's joint.
about 250 years. The alternative was the flexible coupling in

which a leather, fabric or rubber disk was used to transmit
rotary motion through a very limited angle. For larger angles
the Hooke's joint was always used, but this again was less
satisfactory when the angle between drive and driven shafts
became large. This problem arises from the cyclical speed
variation that occurs in the driven shaft.
When we consider the case of a drive shaft rotating at
constant angular velocity, the driven shaft undergoes a
cyclical speed variation every 180 degrees of rotation, being
slower than the drive shaft at first then faster during the final
90 degrees of rotation. The mean angular velocity over the
180° cycle is, of course, identical to that of the drive shaft.
For an angular shift of 8° between the lines of the two shafts,
the maximum variation in angular velocity is only 2%. If we
treble the angle between the shafts to 24°, the maximum
velocity variation increases to 18%. The standard technique of
coping with this problem is to use two Hooke's joints, 90°
Fig. 11.2 The Rzeppa constant-velocity universal joint.

The Fiesta suspension 187
out-of-phase, as shown in Fig. 11.1. The intermediate shaft
still undergoes cyclical variations, these can produce
undesirable vibrations that can be transmitted to the body if
shaft angles of 20° are exceeded. With FWD much greater
angles than 20° are required, if acceptable turning circles are
to be given.
The need for a better VII was apparent for more than a
generation and three engineers, Gregoire, Rzeppa and Weiss,
all working along similar lines eventually developed
constant-velocity universal joints that were reliable and
compact. A typical Rzeppa joint is shown in cross section in
Fig. 11. 2. Six hardened steel balls, rolling in grooved
raceways in the two halves of the joint, transmit torque across
the coupling. The six balls are located by a cage and the
purpose of the cage is to maintain the six balls in the same
plane, this plane being at an angle, the half angle 1X12, where
IX is the relative angle between drive and driven shafts. As the
angle IX changes, so does the operating plane of the six balls.
By maintaining the driving members (the balls) in the median
plane, torque remains constant and the drive is therefore at
constant velocity. There are variants on this basic design,
usually in the profile of the ball tracks. Peugeot now use what
is called a 'tripod joint', in which three hardened rollers
transmit the drive. The principle is identical to that used by
Rzeppa and it also gives constant velocity. Some designs have
been developed to accept 'plunge'. In these joints provision is
made for lateral movement in a sliding sleeve, since the
suspension geometry sometimes calls for a small change in
drive shaft effective length.
11.3 The Fiesta suspension
Today, very few cars are designed by one man. Even cars
that are made in half-dozen batches like Formula 1 racing cars
have a specialist like Keith Duckworth to design the engine.
Again, Hewland Engineering Ltd design the transaxle and an
overall chassis specialist like Derek Gardner designs the rest.
When a mass-produced car such as the Ford Fiesta is
designed, relatively large specialist teams are involved with a
project leader to co-ordinate every part of the project with the
188 A small FWD saloon car: Fora Fiesta S
whole. Inevitably, committee efforts are slow but the vast
inveshnent involved encourages caution. The executive
engineer in charge of chassis design on the Fiesta project was
Tony Piilmer and his original plan for a wishbone suspension
system at the front using torsion bars was changed to the
MacPherson strut system, a design used on several earlier
British Ford cars.
11. 3.1 The front suspension

The general arrangement of the front suspension is shown in
Fig. 11.3; while a cross section of the wheel bearing assembly
is given in Fig. 11. 4. Two aims were dominant in the plans
for the Fiesta: ease of assembly and ease of maintenance. The
first would benefit both supplier and customer; the second
would be very welcome to the customer. A good example of
this philosophy is seen in the front wheel bearings. The
Fiesta uses Timkin '5etright' tapered roller bearings. These
require no adjushnent during assembly, nor the use of an
Fig. 11.3 The Fiesta front suspension.

Fig. 11.4 Fiesta front wheel bearing assembly.
assortment of shims of differing thicknesses. These bearings

are installed so that the inner races abut one against the
other. They are not located by the inner diameter. Tight
control on machining tolerances of the hub carrier permits
bearings to be selected at random and to give a preload on
the bearings within acceptable limits when the hub carrier is
torqued on its stub axle shaft. As in most modern designs,
the bearings are prepacked with grease and 'sealed for life'.
This design, then, does appear to meet the two main criteria.
If the bearings should ever need replacement, the elimination
of shims or other preload adjustment should make it a simple
snagfree operation.
The design of the MacPherson strut differs from standard
Ford practice only in the provision of a top bearing. Normally
on British Ford cars rotation of the strut is accommodated by
190 A small FWD saloon car: Ford Fiesta 5
rotation of the spring and by deflection of the rubber mount
that supports the rear spring carrier. To reduce self-aligning
torque, Palmer provided a phenolic resin moulding to act as a
thrust bearing. To minimize friction, the thrust surface is
impregnated with PTFE. Tie rods are provided to give
fore-and-aft location of the bottom ends of the struts and to
resist brake torque. Rubber bushes are provided at the front
of the tie rods and in the transverse arms to give a moderate
amount of compliance.
The steering geometry is designed to give a 8-mm 'negative
offset'. This is an essential safety feature with a dual braking
system designed to provide a diagonal split. With a positive
offset, if one brake circuit fails and the car were braked on
one front wheel and the diagonally opposite rear wheel, there
would be a tendency for the car to steer towards the braked
front wheel. A small negative offset counteracts this tendency.
11.3.2 The rear suspension

Effectively the rear suspension is a dead axle located by three
main links and two torque straps. The tubular axle is cranked
upwards in the middle (see Fig. 11. 5) and is located laterally
by a Panhard rod. Fore-and-aft location is provided by two
channel-section trailing links attached to brackets below the
axle tube. A degree of compliance is given by rubber bushes
at the forward pivots. A full five-link location would require
the provision of two additional longitudinal links, leading or
trailing, attached to brackets above the' axle tube. Such a
system would be required on a modem live rear axle with coil
spring suspension. A live rear axle, however, must be
provided with linkages to resist torque reaction in both
directions, since there will be a tendency to rotate the axle
tube in one direction under drive torque and in the other
direction under braking torque. The Fiesta rear axle will only
tend to twist under braking torque and this has been resisted
in Palmer's design very simply by the provision of short
straps welded to the lower end of each damper tube and
extended forwards above the axle tube. Rubber bushes are
used to give some compliance where these straps are attached
to vertical pins welded to brackets forward of the axle tube.
Fig. 11.5 The Fiesta S rear suspension.
Under braking forces axle windup is thus transferred through

the straps to the dampers. Piston rods of a diameter larger
than would nonnaUy be used have been provided to resist
this side load. An anti-roll bar is fitted at the rear, but only
on the S model.
Suspension data: Ford Fiesta S
Wheelbase, 2.286 m
Front track, 1.334 m
Rear track, 1.321 m
Kerb weight, 727 kg
Laden weight (4-up), 1040 kg
Sprung mass (4-up) (estimated), 945 kg
Front/rear weight distribution (4-up), 54/46
Tyre size, 145-12
Front spring rate, at wheel, 21.7 kN/m
Rear spring rate, * at wheel, 25.0 kN/m
Rear roll bar rate, at wheel, 3.3 kN/m
Total wheel travel, front, 143mm
Total wheel travel, rear, 166mm
Wheel travel in bump, front, 62mm
Wheel travel in bump, rear, 84mm
Height of roll centre above ground, front, 185mm
Height of roll centre above ground, rear, 191 mm
Roll stiffness, front, 296 Nm/deg
Roll stiffness, rear, 210 Nm/deg
Roll angle at a lateral acceleration of O. 5g, 3° 54'
Camber angle, static, at kerb weight, 1 ° 57' positive
Camber angle, static, laden, 0° 52' positive
Toe setting, 2.5 mm, toe-out
Kingpin inclination, 15° 6'.
Height of sprung mass centroid. Since we know the combined

roll stiffness at front and rear, and we know the roll angle
produced by a lateral acceleration of 0.5 gravity we can
calculate the height of the centroid:
Total roll stiffness = 296 + 210 = 506 Nm/deg.

Roll resistance at 3° 54' (3.9°) = 3.9 x 506 = 1973 Nm.
The roll moment (see Fig. 11.6) = F x h1 = 0.5g xM x h1
= 0.5 x 9.807 x 945 x h1
= 4634h 1 •
Therefore
h = 1973
1 4634
= 0.426 m.
185 + 191
Mean roll centre height = 2 = 188 mm.
Therefore centroid height a hove ground = 426 + "188 = 614 mm.
• This rate is based on both rear wheels in bump and rebound. Single
wheel rate is 12% higher.
Fig. 11.6
11.3.3 Suspension balance

The following data from Chapter 3 will help us to analyse the
overall balance of the Fiesta S suspension in bounce and
pitch:
One-up Four-up
Front spring frequency, in bounce, 1.55 Hz 1.45 Hz
Rear spring frequency, in bounce, 2.17 Hz 1.69 Hz
Pitch frequency, 1.96 Hz 1.56 Hz
Front wheel distance from centroid 11 0.846 m 1.052 m
Rear wheel distance from centroid 12 1.440 m 1.234 m
Front conjugate point distance behind
centroid s 1.440 m 1.074 m
Rear conjugate point distance in front
of centroid r 0.846 m 1.269 m
The dimensions r, S, 11 and 12 are shown in Fig. 11. 7. The
significance of the double conjugate points rand S was
discussed in Chapter 3. On the Fiesta with only the driver in
the car the double conjugate points coincide exactly with the
wheel centres. This is obviously the designer's choice and
means that for this particular loading each 'axle' has no effect
Rear Front
conjugate conjugate
point point
--~~--_Il--~~.~!
..I. s~
Fig. 11.7
in bounce or rebound on the other 'axle'. Analysis of several

modern cars shows this to be a popular choice. As the load is
increased, the rear conjugate points moves forward slightly
and the front conjugate point moves ahead of the rear axle.
Thus any bounce movement at the rear will produce a small
amount of bounce at the front in the same direction. Since the
front conjugate point with four-up falls inside the wheelbase,
any bounce at the front will produce a small amount of
bounce at the rear, this time in the opposite direction. The
change from the one-up condition is not great. A bump
movement of 40 mm at the front would only result in a
re bound movement at the rear of 3 mm. The most significant
change from the one-up load to the four-up load is the
marked reduction in the rear spring frequency and in the
pitch frequency
11.3.4 Roll balance

It has already been stressed that a small family car is subject
to a fairly wide variation in load distribution. It is, therefore,
difficult to balance the roll resistance under all loads. The roll
resistance on the Fiesta S is 296 Nm/deg at the front and
159 Nm/deg from the coil springs at the rear plus 51 Nm/deg
from the anti-roll bar. Expressed as a percentage, the total roll
resistance is split 58.5% at the front and 41.5% at the rear.
This is a very good compromise between the weight
distributions of one-up (63/37) and four-up (54/46).
The FWD cornering syndrome 195
The reader may wonder how the two front springs of
21. 7 kN/m rate can exert a greater roll resistance than two
rear springs of 25.0 kN/m rate assisted by an anti-roll bar
with an effective rate at the wheel of 3.3 kN/m. The answer
lies in the effectiveness of the springs in roll, the beam axle
being less effective. The roll resistance of a beam axle is
proportional to the square of the spring base, whereas the roll
resistance of an independently sprung vehicle is proportional
to the square of the wheelbase. The rear spring base (the
distance between the coil spring centres) is only 72% of the
wheelbase. Compared with the front suspension, the rear
springs are therefore little more than half as effective in
controlling roll.
11.4 The FWD cornering syndrome
In Chapter 6 we illustrated by vector diagrams how an FWD

car can, by accelerating through a curve, comer at a higher
speed than an RWD car. The danger that could exist when
cornering at the limit if the designer has not designed a safety
factor into the suspension and steering geometry has also
been pointed out. Without careful design, the driver could
find himself at the point of no return, where the need to lift
his right foot will inevitably result in a spin. One factor on
the little Fiesta reduces this danger to some extent: there is
very little torque available to spare to exploit the acceleration
technique when cornering at speed. If we refer back to Fig.
6.1(b) and (c), we see that the driver of an FWD car can use
some of the available torque from the driven wheels to direct
the car inwards against the centrifugal force, a technique not
available to the driver of an RWD car.
Ever since General Motors were 'persuaded' by Ralph
Nader to withdraw the original Corvair from production on
the grounds of unsafe behaviour when cornering when driven
by a driver of average ability, every senior engineer responsible
for the production of a new car has been made aware of his
ultimate responsibility. The president of the company may
say to the Press 'the buck stops here', but the senior design
engineer in charge of a new project knows whose job is really
at stake! Ironically, a few modifications to the Corvair
suspension geometry had already made the car perfectly safe,
but withdrawal of the model was advisable after so much
adverse publicity.
The problem of power liftoff on the FWD car is a different
one, but it cannot be ignored even on a low-powered car.
How then can a suspension engineer build a failsafe feature
into the cornering behaviour of an FWD car? When cornering
at the limit, a need to slow down for an obstruction ahead
could result in a breakaway at the front end. What is needed
is a geometry that gives understeer in a corner, but a
reduction in understeer (reduced slip angles at the front or
increased slip angles at the rear) when the car is decelerated
in a bend. There are several ploys available, and ingenious
engineers (the adjective and the noun are from the same Latin
root) prove their worth every few months by inventing new
ones. One recent example from the Porsche engineers is the
Weissach axle, described in Chapter 12. This is a suspension
design for an RWD car which changes from neutral steer to
understeer under liftoff.
Two factors in the steering and suspension geometry of the
Fiesta tend to give understeer when under power in a bend
and oversteer when the foot is lifted from the accelerator, or
even under a moderate brake application. The first factor is
the position of the steering rack. If the steering rack is
mounted behind the front wheels, lateral forces cause the
bushes to deflect in such a way that a small amount of
oversteer is produced. This is the posit jon of the steering rack
on the Fiesta. Roll understeer can be designed into a trailing
link located rear axle by the inclination of the links as shown
(much exaggerated) in Fig. 6.6. With the forward pivots of
the trailing links below wheel hub height, the outer wheel
moves forward when rolling in a bend and the inner wheel
moves backward: this gives understeer.
Let us consider then what happens when acceleration in a
fast bend is suddenly followed by deceleration. For stable
cornering, the balance between front and rear tyre forces and
roll moments have been carefully balanced. Front slip angles
will be a little greater than those at the rear. As soon as the
brakes are applied (even a driver of average ability should
know not to stamp on them in the middle of a bend), there is
a transfer of load from the rear to the front. This increases the
The FWD cornering syndrome 197
lateral forces on the steering bushes, and with sufficient
compliance this oversteering effect reduces the overall
understeer. The second effect of braking is that the reduction
in load on the rear axle gives a reduction in the rear roll
angle. This again gives a reduction in understeer. It is
doubtful if the combined effect of these two factors would
make any FWD car completely 'c1otproof' on a slippery bend,
but the experience of the motoring Press, with many miles of
testing in the severe British winter of 1978-9, has been
unanimous. The handling of the Fiesta is safe and
predictable.
12
A High-performance Sports Car:
Porsche 928
12.1 Porsche panache
The Porsche Co. has a long history of development in the

field of suspension and handling, some of this won the hard
way by attempting to solve problems introduced by the
inherent weaknesses in their initial design. *
Many readers will be familiar with the Porsche 911, that
brilliant descendant of a long line of sports cars with
air-cooled engines at the rear. The evolution of semi-trailing
link suspensions at both front and rear has been finely tuned
to give excellent handling.
The new models emanating from the Porsche Design
Office, the 2-litre 924 and the 4.5 litre 928, are most
surprisingly front-engined cars with rear wheel drive. The
final shock is the switch from air-cooled to water-cooled
engines, but this is a subject outside the scope of this book.
Helmut Bott, the Director of Engineering at Porsche, has,
stated that the change to a front-engined location was made
to ease the problems associated with the USA 30 m. p.h.
simulated front-end collision test. By using a rigid tube to
connect the engine and rear-mounted transaxle, the impact
energy can be absorbed by the main mechanical components.
Fig. 12.1 is a ghosted view of the Porsche 928. The engine
central tube and transaxle can be seen in the photographic
plan in Fig. 12.2.
• Dr Porsche fought doggedly with the inherent handling problems of swing

axle suspension. On the 520 b.h.p. Auto Union P-Wagen GP car the
problems were almost insuperable.
200 A high-performance sports car: Porsche 928
Fig. 12.2 Porsche 928 with body removed.
12.2 The Porsche 928 suspension
Upper and lower wishbones are used at the front. At the rear
upper and lower transverse links are used with the addition
of a special patented flexibly mounted trailing link known as
the Weissach-Axle. The suspension geometry gives 30%
anti-dive at the front and 50% at the rear. The rear
suspension layout also gives 70% anti-squat. Compromise is
essential in these factors, especially when designing a sports
car, if adverse changes in toe-in, camber ~nd castor angles are
to be avoided. The suspension engineers at Weissach, the
Porsche research centre, admit that they arrived at the chosen
geometry by trial and error. With a new untried design one
can hardly expect the computer to give a reliable answer.
On the front suspension an alloy casting is used to provide
a conventional upper wishbone. An even more robust alloy
casting is used for the transverse lower arm. These
components are particularly robust since they are subjected to
very high forces during braking, acceleration and cornering.
'By comparison the suspension components on the Fiesta
are quite flimsy. The Porsche 928, however, has a
power-to-weight ratio of 115 kW/tonne (160 b.h.p./ton) and
the Pirelli P7 225/50 VR16 tyres, fitted as standard, endow
The Porsche 928 suspension 201
the car with phenomenal cornering accelerations. The use of
tyres with such a low-profile ratio does, however, involve the
penalty of a suspension geometry that allows only small
changes in wheel camber. At the front the variation from full
bump to full rebound is only about 1.5°. At the rear it is less
satisfactory since it approaches 4 degrees.
The components of the rear suspension can be seen in Fig.
12.3, which is a photograph of the rear suspension
cross-member and the suspension components taken from the
front. Each wheel hub carrier is located by a simple
transverse upper link and a more complex lower link which
is, in effect, an integrated two-part link. The transverse
member is a relatively thin steel plate, very stiff under pure
vertical loads, but flexible in the fore-and-aft direction. The
trailing link is attached to an articulated double-pivot, which
lies approximately in line with the inner pivot (the pivot of
the steel plate) to make an angle of approximately 25° with
the centre line of the car. The Weissach-Axle is, therefore, a
semi-trailing link suspension with a variable axis. The
dangers of excessive amounts of compliance in the rubber
bushes used in suspension links are well known to
suspension engineers. In racing car design, where small
variations in suspension geometry cannot be tolerated, rubber
bushes are never used. Drivers of ordinary passenger vehicles
do not expect to suffer shocks, vibrations or noise transmitted
directly from the suspension attachments. Isolation, within
Fig. 12.3 Porsche rear suspension, viewed from the front.

ex
Decelerating
(a)
Deceferating
( b) (c)
Fig. 12.4 (a), Porsche 928 conventional suspension toe-out due to
elasticity. When decelerating or braking, the wheels in a
conventional suspension toe-out at a angle owing to force P. The
reasons are the rubber bearings in the suspension required for noise
absorption. (b)-(c), Weissach-Axle - toe-in due to kinematics. A
kinematic effect changes the angle towards toe-in when decelerating
or braking. For reasons of nOIse absorption, rubber bearings are
featured, but toe-out is immediately balanced by the kinematic effect
a = p. Thus dangerous side movements are prevented in curves, an
important characteristic.
the limits of current technology, is therefore the order of the
day, even on a sports car.
Cunning old Americans say 'If you can't lick 'em, join 'em'.
This is the philosophy of the Weissach-Axle. Rubber
mountings are used on the Porsche 928 to modify suspension
behaviour under traction, free roll and deceleration to give an
automatic correction to the amount of rear wheel toe-in. Even
when cornering near the limit, the fine tuning of the variation
in toe-in calls for negligible steering correction from the driver
when forced to decelerate.
With greatly exaggerated angles, the Weissach-Axle
principle is explained in Fig. 12.4. On a conventional
semi-trailing link rear suspension deceleration (and to a
greater extent, braking) deflects the rubber-mounted pivots
to give a toe-out angle to both rear wheels. This is an
oversteering effect. Since braking also robs the rear tyres of
effective cornering force and increases the rear slip angles (see
the Circle of Forces, Fig. 1.17), these two oversteering effects
in combination can sometimes give the driver a difficult
handling problem when decelerating in a corner. To the
professional test driver, this is often termed liftoff tuck-in.
When very rapid unwinding of the steering wheel is
necessary to correct this oversteer, the car can hardly be
called 'inherently stable'.
The front pivot of the Weissach-Axle is shown
schematically in Fig. 12.4. It consists of two bushes, a
conventional rubber bush on a rod end and a larger rubber
bush, 100 mm long and 50 mm diameter, to which the rod
end is fixed. Only very small changes in the amount of toe-in
are permitted by this double-jointed pivot system. The static
toe-in is 20', decreasing under traction, but never reaching a
toe-out angle. Under severe braking, the toe-in will increase
but will seldom reach more than 50'. These are very small
angular changes, but sufficient to give vicefree cornering
behaviour.
The following data have been supplied by Porsche:
Type 928 chassis
Wheelbase, 2.5m
Track, front, 1.545 m
Track, rear, 1.514 m
Kerb weight, 1450 kg
Permissible total weight, 1870 kg
Weight distribution, one-up, 49/51
Weight distribution, two-up plus luggage, 47/53
Type 928 suspension

Front Rear
Wheel rate, 18.63 kN/m 22.55 kN/m
Anti-roll bar rate, at wheel, 83.4 kN/m 10.2 kN/m
Total roll stiffness, 2115 Nm/deg 655 Nm/deg
Wheel travel, bump, 90mm 100mm
Wheel travel, rebound, 90mm 100mm
Roll centre height above
ground, 70mm 110mm
Camber angles
full bump, _2° 8'
normal laden, -30'
full re bound, -42'
Castor angle, 3° 30'
Kingpin inclination, 13° 36'
Toe-in, static, o 20'
12.2.1 Yaw response

Modern wide-section tyres have exerted a profound influence
on the thinking of designers of sports cars. Twenty years ago
one could argue with no thought of contradiction, that a
sports car should have a low polar moment of inertia, this
being the polar moment about the vertical axis as discussed
in Chapter 6. At that time it was contended that a rapid
change of direction, such as a lane-change on a motorway or
a chicane on a race circuit, could be more accurately and
safely executed when the polar moment of inertia was low.
Today, the grip of modern tyres with their ultra-low profiles
and much-improved rubber compounds makes it difficult to
establish which sports cars have the least tendency to spin,
since very rapid changes of direction can be made even in the
wet by sports cars with both high and low polar moments of
inertia. Experiments were carried out in 1977 (Motor, May
7th 1977) on a range of sports cars with polar moments of
inertia (measured at the Cranfield Institute of Technology)
varying between 1112 kg/m 2 for the little mid-engined Fiat
Xl/9 to 2050 kg/m2 for the Porsche Carrera. In these
experiments handbrakes were applied to lock the rear wheels
then released immediately, while the car was in the middle of
a 50-m radius tum. The wetted road surface had a coefficient
of friction of approximately 0.3. The curve was negotiated at
a speed of 30 m. p.h. This technique (reminiscent of the
hand brake tactics used by rally drivers when they require to
put their car sideways-on as they approach a tight tum)
inevitably made the rear end of every sports car tested
breakaway and begin to spin. The cars were all equipped
with instruments to measure the yaw acceleration, since it was
reasonable to suppose that a car that spun with the highest
rate of angular acceleration would be the one that would be
the most difficult to control by steering correction.
The experiments were inconclusive. No correlation was
apparent between maximum yaw accelerations and polar
moments of inertia. The 'seat of the pants' impressions of the
drivers gave no indication that the tendency to spin on a
slippery surface was any greater with front-engined or
rear-engined cars or with those of low or high polar moment
of inertia. It is not surprising, therefore, that the Porsche
design team working on the type 928 had no hesitation in
adopting a completely different layout of the major masses
than they had used in the past. Since they planned the
chassis on the basis of identical low-profile tyres at front and
rear, they saw that the essential requirement was a mass
distribution that gave almost equal static tyre loads. This has
been achieved with a 49/51 weight distribution with a one-up
load and a 47/53 distribution with two-up plus luggage.
With the phenomenal grip offered by the Pirelli P7 tyres the
second essential requirement was a close control on camber
angle change. This is achieved on the front suspension with a
variation between full bump and full rebound of only 1.5
degrees. The camber variation at the rear is greater, varying
between - 4 degrees 30 minutes at full bump and - 33
minutes at full rebound. This represents the maximum change
of camber between bump and rebound stops and is no
measure of the camber changes experienced in roll.
12.2.2 Roll
Roll is controlled on the Porsche 928 by the provision of stiff
anti-roll bars, in particular at the front. Since the roll centre is
40 nun lower at the front than at the rear (see Suspension
Data, above), one would anticipate the need for a higher roll
resistance at the front. The roll resistance at the front is
actually about three times that at the rear.
The combined roll stiffness of suspension springs plus both
anti-roll bars is 2770 Nm/deg. We have taken an estimated
value of the sprung mass centroid height at a value of
440 mm. Using this, we can make an estimated approach to
the roll angle under a chosen lateral acceleration.
Using the same nomenclature as in Chapter 11 (see Fig.
11.6):
h 2 (mean) 90 mm
hI (estimated) = 440 - 90 = 350 mm
The roll moment = F x h 1.
At 1 gravity centrifugal acceleration, the roll moment
RM = 1.0 x 9.807 x 1385 x 0.350 = 4754 Nm
where 1385 kg is the estimated sprung mass (one-up). Since
the roll stiffness is 2770 Nm/deg, the mean roll angle at 1
gravity cornering acceleration is:
4754 = 1. 7° = 1 42'.
0
2770
It is interesting to compare this with the Ford Fiesta S in
Chapter 10 which rolls through 3°54' under 0.5 gravity
centrifugal acceleration.
12.2.3 Camber change in roll

A roll of 1 42' on the type 928 will give a bump movement
0
on the outer rear wheel of sin(l °42') x 757 mm, where

757 nun is half the rear track.
This is a bump movement of 22.7 nun (about one-quarter
of the available bump travel). This will increase the normal
laden camber angle at the rear from _1 to about -1°48'.
0
The camber changes on the inner rear wheel and the two
front wheels will be negligible. Thus by the provision of good
Natural frequencies and double conjugate points 207
roll resistance, Porsche have met this essential requirement for
a car fitted with 50 series tyres.
12.3 Natural frequencies and double conjugate points
Using Professor Guest's method as outlined in Chapter 3, we

have sufficient data to make a close approximation to the
natural suspension frequencies and double conjugate points
with different laded weights:
Frequency data (one-up)
Sprung mass Ms (estimated) = 1385 kg
Wheelbase = 2.5 m
Weight distribution = 49/51
II = 0.51 x 2.5 = 1.275 m
12 = 0.49 x 2.5 = 1.225 m
kl = 18.63 kN/m
k2 = 22.55 kN/m
k2
a =k-k x2.5 = 1.369 m
1 + 2
kl
b =k-k x2.S = 1.131 m
1 + 2
K2
I I (estimated*) = 0.98
1 x 2
Therefore
K2 = 1.5306
c2 = a x b = 1.548
x = a -II = 0.094 m
x2 = 0.0088.
From this data we can obtain the following values:
Conjugate Points (one-up)
Front conjugate point distance behind centroid s = 1. 382 m
Rear conjugate point distance in front of centroid r = 1.103 m.
* For the type 924 (the 2-litre model) Cranfield Institute measured a value
of 0.95.
Natural frequencies (one-up)
Ff = 1.18 Hz
Fr = 1.28 Hz
F pitch = 1.24 Hz.
Frequency data (two-up plus luggage)

Sprung massMs (estimated) = 1660 kg
Wheelbase = 2.5 m
Weight distribution = 47/53
11 = 0.53 x 2.5 = 1.325 m
12 = 0.47 x 2.5 = 1.175 m
k1 = 18.63 kN/m
k2 = 22.55 kN/m
K2 = 1.5306
a = 1.369 m
b = 1.131 m
c2 = 1.548
x = a - l1 = 0.044 m
= 0.001936.
This data for the more heavily loaded condition gave the
following values:
Conjugate points (two-up plus luggage)

s = 1.476 m
r = 1.037 m.
Natural frequencies (two-up plus luggage)

F f = 1.10 Hz
Fr = 1.14 Hz
Fp = 1.13 Hz.
12.4 Comments
For a sports car the natural frequencies of the suspension,

both front and rear, are exceptionally low. Over the majority
Comments 209
of road surfaces the ride cannot fail to be good, with much
lower vertical accelerations than one finds in many modem
sports cars. Since Porsche have chosen to make front and rear
natural frequencies almost identical, it is inevitable that the
natural frequency in pitch is also very close to these values.
With much higher frequencies - as, for example, those used
for the Ford Fiesta - it would be inadvisable to permit such a
close approach, but the use of a well-damped suspension in
the Porsche has reduced the dangers of pitch and bounce
movements overlapping to give any discordant 'joggle' to the
ride.
One penalty associated with the use of such a stiff anti-roll
bar as that used at the front of the Porsche 928, is that the
spring rate for single wheel bump is much higher than that
for double wheel bump. The frequency in single wheel bump
is controlled by three springs. Of course, the first spring is
the main spring on the side involved. Added to this is the
spring rate of the anti-roll bar acting in series with the main
spring on the other side of the car.
As discussed in <,::hapter 4, two springs working in series
behave as if replaced by an equivalent spring S e' where
The normal wheel rate at the front is 18.63 kN/m and the
anti-roll bar rate is 83.4 kN/m. The additional equivalent
spring rate is therefore:
S = 83.4 x 18.63 = 15.23 kN/m.

e 83.4 + 18.63
The total spring rate for single wheel bump at the front is
therefore 18.63 + 15.23 = 33.86 kN/m.
Since the natural frequency of a spring system varies as the
square root of the effective spring rate, the natural frequency
for single wheel bump at the front is 35% higher than the
frequency for double wheel bump. The rear anti-roll bar is
not very stiff and the single wheel bump frequency at the rear
is only about 14% higher than the double wheel bump
frequency.
The noticeable hardening of the front suspension under
single wheel bump has received some adverse comments
from a few members of the motoring press. On the whole, it
appears to be a small price to pay for a car that rides so well
and comers like a real sports car - and that undoubtedly is
the best description of the Porsche 928.
Index
Acceleration, 20, 21, 109, 192 Chapman strut, 66, 69

angular, 113, 116 Circle of Forces, 21, 108, 203
vertical, of car body, 25-27 Citroen 'hydropneumatic suspension,
of human body, 22-23, 26-27 143-158
Ackermann steering, 61-62, 106 2CV interconnected suspension,
Active suspension, 180-183 166-167
Aerodynamic forces, 110 GS front suspension, 152-154
Air spring, 129-132 rear suspension, 154-155
Anti-dive, 72-73, 157, 164 Comfort, 22-23, 40
Anti-roll bar, 70-72, 154-155, 194, Conjugate points, 45-52
195, 206, 209-210 Constant velocity VI], 185-187
height control system, 151-152 Cornering dynamics, 106-112
Anti-squat, 72, 75-76, 156, 164 force, 17, 18, 20, 21, 60, 108, 175,
Aquaplaning, 15-16 176
Aston Martin rear suspension, 85 power, 13, 18, 19, 20, 21, 60,
Austin Allegro suspension, 173 108, 175, 176
Automatic Products no-roll Coupled suspensions, 41-53
suspension, 181-183
Axle, beam, 64, 77-79 Damper, double-tube, 122, 123
dead rear, 99 friction, 120
de Dion, 65, 84-85, 175 Girling, 122-124
live rear, 65, 79-84 hydraulic, 120-122
swing, 65-66, 67, 68 Koni, 122, 123
effective length, 104 VVoodhead-Munroe, 124-125
Damping theory, 125-128
Banking car, 173-174 Dead rear axle, see Axle, dead rear
Beam axle, see Axle, beam de Dion suspension, see Axle, de
Bounce, 42, 49, 52, 53 Dion
frequency, 52, 53, 133, 134, 153, Deflection, static, 27, 28, 29, 30
172, 173, 192, 193, 208 steer, 118-119
period, 44, 50 Directional stability, 56-57, 105-106
Braking, 20, 21, 72, 73, 74, 75, 156, Double conjugate points, 47-57
157, 158, 196, 197, 202, 203 Dynamically conjugate points, 46-47
Bump, 27, 28, 88, 92, 96, 103, 147,
151, 153, 154, 164, 166, 167,
170, 205, 206, 209, 210 Effective swing arm centre, 103, 104
Elastically conjugate points, 45-46
Camber, 19-20, 102, 103, 117, 205,
206 Fatigue, 22-23
angle, 28, 54, 55, 56-57, 88, 104, Firestone air suspension, 132-138
177, 192, 201, 204, 206 Footprint, see Tyre footprint
Castor angle, 54, 55, 60 Ford Fiesta
Centrifugal force, 108, 111, 114, 177 roll balance, 194-195
212 Index
suspension, 51, 187-197 Negative offset, 59-60, 97-98, 190
balance, 193-194 No-roll suspension, 93-96, 173-183
Granada rear suspension, 104
Front wheel orientation, 54-62 Over centre-point steering, 97, 98
FWD, 185-187 Oversteer, 57, 105, 106, 119, 203
cornering, 109-110, 195-197
Passive suspension, 180
General Motors air suspension, Periodic time, of SHM, 25
138-141 about double conjugate points,
Ground effect car, 110, 175, 177 50-51
Gyration, radius of, 43, 44 Pitch, 42, 43, 49, 73, 165, 166, 167,
Gyroscopic torque, 78 169, 172, 209
frequency, 52, 53, 167, 173, 193,
Hydragas suspension, see Moulton 208
suspension period, 44, 51
Hydraplaning, 16 rate, 172, 173
Hydrolastic suspension, see Moulton Pneumatic spring, see Air spring
suspension Polar moment of inertia, see Inertia,
Hysteresis, tyre, 36 polar moment of
Porsche, Professor, 65, 88
Independent front suspension, 77, swing axle, 65-66
85-98 trailing link suspension, 65, 88-89
rear suspension, 98-104 Type 911 suspension, 198
Inertia, moment of, 42 928 roll behaviour, 206
polar moment of, 43, 46, 50, 51 spring frequencies, 207-208
Interconnected front/rear suspension, suspension, 198-210
165-173 Weissach-Axle, 201-203
yaw response, 204-205
Jacking action, 66, 67
Jaguar rear suspension, 100-101 R-Type MG suspension, 90-91
Jounce, 28 Radius of gyration, 43, 44
Rebound, 28, 88, 92, 96, 103, 151,
Kingpin inclination, 55, 57-60 153, 154, 164, 166, 167, 170,
offset, 55, 59, 97 205
Resonance, suspension, 33-34, 127
Leading arm, 73, 74, 75, 157 Ripples, road, influence of, 29-33
Lift off tuck-in, 203 Roll angle, 88, 91, 93, 95-96, 176,
Live rear axle, see Axle, live rear 192, 206
Lotus 79 cornering analysis, 175-177 axis, 69-70
centre, 62, 63, 64, 65, 66, 69, 70,
MacPherson strut, 66, 69, 96-98, 175, 176, 177, 179
189-190 geometry, 62-72
Mercedes-Benz hydro pneumatic moment, 63, 175, 176, 192, 206
suspension, 158-164 physiology, 94-96
MG, R-Type suspension, 90-91 steer, 66, 104, 117
Michelin X tyre, 10-13 Rover 3500 rear suspension, 83-84
Moment of inertia, see Inertia,
moment of Self-aligning torque, 59-60
Morgan front suspension, 85, 87, 88 Semi-trailing arm, 102-104, 201, 213
Moulton suspension, hydragas, Shimmy, 54, 77-78
170-173 Simple Harmonic Motion, 24-25
hydrolastic, 167-169 Skidding, 13-16
Index 213
Slide, 114 construction, 9-13
Sliding pillar suspension, 65, 85, 87, contact patch, 12, 29
88 cornering force, 17, 18, 20, 21,
Slip angle, 16, 17, 18, 20, 60, 62, 175-177
105,106,114,196 power, 18, 155
Spin, 114 cross-ply, 10
acceleration, 116-117 footprint, 11, 12, 13, 16, 17, 18
Spring, coil, 27 hysteresis, 36
design, 23-25 motion transmissibility, 40
frequency, 25, 28, 29, 30, 31, profile, 18-19
49-53, 153 radial, 10-13
of the tyre, 34-40 stiffness, 37, 39
laminated leaf, 27
rates, 26, 27-28, 153 Understeer, 105-106, 119, 196, 197
semi-elliptic, 64 Universal joint, see Constant velocity
static deflection, see Deflection,
static U/J
Unsprung mass, 29, 32
stiffness, 25
torsion bar, 27
Sprung mass, 29, 30 Volkswagen Beetle, 88, 114
movement of, 31-33
Straight-line stability, see Directional Weissach-Axle, 201-203
stability Wheel, 1--{'
Suspension theory, 44-51 aluminium alloy, 3
Swing axle, see Axle, swing contact, 29
Swivel angle, see Kingpin inclination magnesium alloy, 4-5
steel, 2-3
Toe-in, 54, 103, 118, 203, 204 wire, 5--{'
Toe-out, 119, 192, 203 Wishbone suspension, angled,
Torque tube, 83-84 91-93, 95, 96
Traction, 20, 21 parallel, 64--{'5, 90-91
avant, 143 rear, 100-102
Trailing arm suspension, 65, 73, 75, roll centre determination, 63--{'4
88-89,157
Tramp, 78-79
Trebron suspension, 177-180 Yaw angle, 106, 108, 111
Tyre, 7-21 inertia, 112-114, 115
bias-ply, 10 response, 115, 116

(Colin Campbell M.SC., C.Eng., M.I.Mech - Engrs PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(Colin Campbell M.SC., C.Eng., M.I.Mech - Engrs PDF

Uploaded by

Copyright:

Available Formats

Automo bile Suspensions

ISBN-13: 978-0-412-16420-0 (cased edition)

British Library Cataloguing in Publication Data

Automo bile suspensions

This book is an introduction to the elementary technology of

The term suspension suggests 'supporting from above', as in

1.2 The wheel

Nobody knows who invented the wheel. That delightful

1.3 Modem automobile wheels

1. 3.1 Steel wheels

~ Rim width ---l

requirements. The flange profile, indicated by letters K, JK, J

1;3.2 Aluminium alloy wheels

Fig. 1.2 Minilite aluminium alloy wheel.

1.3.3 Magnesium alloy wheels

Shadow Fl F3 Rondel F2 Brabham F3

Fig. 1.3 Cross sections of several MEL MAG magnesium alloy

expected to survive with a rim thickness of no more than

1. 3.4 Wire wheels

those of us of advancing years the blood is still stirred by

Fig. 1.5 Thomson's original pneumatic tyre of 1845.

1.4 The tyre;

When the process of vulcanizing natural rubber became a

Fig. 1.6 C. K. Welch's rim and bead design of 1890.

'Later reregistered as Dunlop Pneumatic Tyre Co.

Fig. 1.7 William Bartlett's rim and bead design of 1890.

Two inventions that improved the reliability and

1.4.1 Tyre construction

Fig. 1.8 Internal construction of cross-ply tyre.

Fig. 1.9 Tyre footprints.

'" DUNLOP SP SPORT

Fig. 1.10 Internal construction of steel-braced radial tyre.

Fig. 1.11 Tread patterns of the 1920s.

SLIDING VELOCITY (MPH)

Machine in the 1950s tread patterns were developed and

1.4.3 Dynamic footprint behaviour

Effective Swivel Point

Fig. 1.14 Distortion of tyre as it approaches footprint zone (extent

1° 2° 3° 4° 5° 6° 7° SO 9° 10° 1,° 12°

1.4.4 Cornering power

Tyre profile. The tyre profile ratio is usually expressed as an

Camber. Wheel camber is the angle between the plane of

Fig. 1.16 Camber.

Load. Every tyre is designed for a specified load. There is a

Traction, braking and acceleration. It has been shown

CIRC LE DEFINES LI M ITING

Fig. 1.17 Circle of Forces: how available cornering force is reduced

forces (cornering plus traction, or cornering plus braking).

[1] Setright, L. J. K. (1972), Automobile Tyres, Chapman & Hall,

2.1 Comfort and fatigue

The degree of discomfort we are prepared to accept from an

The medical profession has been interested in the effects of

2.3 Spring design

2.3.1 The simple undamped spring

A simple spring system is shown in Fig. 2.1. In the mean

o;,~o~J pos",o,__ L____._

Fig. 2.1 Simple spring system.

Fig. 2.2 Simple Harmonic Motion.

In the displaced position the body of mass M, if released,

OJ is the angular speed in Circular Harmonic Motion, as

2.3.2 Body accelerations

Taking a typical front spring frequency of 1.25 Hz, i.e.

This maximum acceleration to which the occupants will be

o;,~o~J pos",o, L__._